forester/java/src/org/forester/application/surfacing.java

   1 // $Id:
   2 //
   3 // FORESTER -- software libraries and applications
   4 // for evolutionary biology research and applications.
   5 //
   6 // Copyright (C) 2008-2009 Christian M. Zmasek
   7 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   8 // All rights reserved
   9 //
  10 // This library is free software; you can redistribute it and/or
  11 // modify it under the terms of the GNU Lesser General Public
  12 // License as published by the Free Software Foundation; either
  13 // version 2.1 of the License, or (at your option) any later version.
  14 //
  15 // This library is distributed in the hope that it will be useful,
  16 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18 // Lesser General Public License for more details.
  19 //
  20 // You should have received a copy of the GNU Lesser General Public
  21 // License along with this library; if not, write to the Free Software
  22 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  23 //
  24 // Contact: phylosoft @ gmail . com
  25 // WWW: www.phylosoft.org/forester
  26
  27 package org.forester.application;
  28
  29 import java.io.BufferedWriter;
  30 import java.io.File;
  31 import java.io.FileWriter;
  32 import java.io.IOException;
  33 import java.io.Writer;
  34 import java.util.ArrayList;
  35 import java.util.Date;
  36 import java.util.HashMap;
  37 import java.util.HashSet;
  38 import java.util.List;
  39 import java.util.Map;
  40 import java.util.Map.Entry;
  41 import java.util.Set;
  42 import java.util.SortedMap;
  43 import java.util.SortedSet;
  44 import java.util.TreeMap;
  45 import java.util.TreeSet;
  46
  47 import org.forester.evoinference.distance.NeighborJoining;
  48 import org.forester.evoinference.matrix.character.CharacterStateMatrix.Format;
  49 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
  50 import org.forester.go.GoId;
  51 import org.forester.go.GoNameSpace;
  52 import org.forester.go.GoTerm;
  53 import org.forester.go.GoUtils;
  54 import org.forester.go.OBOparser;
  55 import org.forester.go.PfamToGoMapping;
  56 import org.forester.go.PfamToGoParser;
  57 import org.forester.io.parsers.HmmscanPerDomainTableParser;
  58 import org.forester.io.parsers.HmmscanPerDomainTableParser.INDIVIDUAL_SCORE_CUTOFF;
  59 import org.forester.io.parsers.util.ParserUtils;
  60 import org.forester.io.writers.PhylogenyWriter;
  61 import org.forester.phylogeny.Phylogeny;
  62 import org.forester.phylogeny.PhylogenyMethods;
  63 import org.forester.phylogeny.PhylogenyNode;
  64 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  65 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  66 import org.forester.protein.BinaryDomainCombination;
  67 import org.forester.protein.Domain;
  68 import org.forester.protein.DomainId;
  69 import org.forester.protein.Protein;
  70 import org.forester.species.BasicSpecies;
  71 import org.forester.species.Species;
  72 import org.forester.surfacing.BasicDomainSimilarityCalculator;
  73 import org.forester.surfacing.BasicGenomeWideCombinableDomains;
  74 import org.forester.surfacing.CombinationsBasedPairwiseDomainSimilarityCalculator;
  75 import org.forester.surfacing.DomainCountsBasedPairwiseSimilarityCalculator;
  76 import org.forester.surfacing.DomainCountsDifferenceUtil;
  77 import org.forester.surfacing.DomainLengthsTable;
  78 import org.forester.surfacing.DomainParsimonyCalculator;
  79 import org.forester.surfacing.DomainSimilarity;
  80 import org.forester.surfacing.DomainSimilarity.DomainSimilarityScoring;
  81 import org.forester.surfacing.DomainSimilarity.DomainSimilaritySortField;
  82 import org.forester.surfacing.DomainSimilarityCalculator;
  83 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  84 import org.forester.surfacing.GenomeWideCombinableDomains;
  85 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  86 import org.forester.surfacing.MappingResults;
  87 import org.forester.surfacing.PairwiseDomainSimilarityCalculator;
  88 import org.forester.surfacing.PairwiseGenomeComparator;
  89 import org.forester.surfacing.PrintableDomainSimilarity;
  90 import org.forester.surfacing.PrintableDomainSimilarity.PRINT_OPTION;
  91 import org.forester.surfacing.ProteinCountsBasedPairwiseDomainSimilarityCalculator;
  92 import org.forester.surfacing.SurfacingUtil;
  93 import org.forester.util.BasicDescriptiveStatistics;
  94 import org.forester.util.BasicTable;
  95 import org.forester.util.BasicTableParser;
  96 import org.forester.util.CommandLineArguments;
  97 import org.forester.util.DescriptiveStatistics;
  98 import org.forester.util.ForesterConstants;
  99 import org.forester.util.ForesterUtil;
 100
 101 public class surfacing {
 102
 103     private static final int                                  MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING                                  = 1000;
 104     public final static String                                DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS                           = "graph_analysis_out";
 105     public final static String                                DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS                       = "_dc.dot";
 106     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS        = "_fitch_present_dc.dot";
 107     public final static String                                DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX                                    = ".dcc";
 108     // gain/loss:
 109     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS                                      = "_dollo_gl_d";
 110     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_BINARY_COMBINATIONS                          = "_dollo_gl_dc";
 111     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_DOMAINS                                      = "_fitch_gl_d";
 112     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS                          = "_fitch_gl_dc";
 113     // gain/loss counts:
 114     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS                               = "_dollo_glc_d";
 115     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_BINARY_COMBINATIONS                   = "_dollo_glc_dc";
 116     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_DOMAINS                               = "_fitch_glc_d";
 117     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS                   = "_fitch_glc_dc";
 118     // tables:
 119     public final static String                                PARSIMONY_OUTPUT_FITCH_GAINS_BC                                               = "_fitch_gains_dc";
 120     public final static String                                PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC                                          = "_fitch_gains_dc.html";
 121     public final static String                                PARSIMONY_OUTPUT_FITCH_LOSSES_BC                                              = "_fitch_losses_dc";
 122     public final static String                                PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC                                         = "_fitch_losses_dc.html";
 123     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_BC                                             = "_fitch_present_dc";
 124     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC                                        = "_fitch_present_dc.html";
 125     public final static String                                PARSIMONY_OUTPUT_DOLLO_GAINS_D                                                = "_dollo_gains_d";
 126     public final static String                                PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D                                           = "_dollo_gains_d.html";
 127     public final static String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_D                                               = "_dollo_losses_d";
 128     public final static String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D                                          = "_dollo_losses_d.html";
 129     public final static String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_D                                              = "_dollo_present_d";
 130     public final static String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D                                         = "_dollo_present_d.html";
 131     public final static String                                DOMAINS_PRESENT_NEXUS                                                         = "_dom.nex";
 132     public final static String                                BDC_PRESENT_NEXUS                                                             = "_dc.nex";
 133     // ---
 134     public final static String                                PRG_NAME                                                                      = "surfacing";
 135     public static final String                                DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                                    = "_d_dollo"
 136                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 137     public static final String                                DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH                                    = "_d_fitch"
 138                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 139     public static final String                                BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                 = "_dc_dollo"
 140                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 141     public static final String                                BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH                 = "_dc_fitch"
 142                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 143     public static final String                                NEXUS_EXTERNAL_DOMAINS                                                        = "_dom.nex";
 144     public static final String                                NEXUS_EXTERNAL_DOMAIN_COMBINATIONS                                            = "_dc.nex";
 145     public static final String                                NEXUS_SECONDARY_FEATURES                                                      = "_secondary_features.nex";
 146     public static final String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES                           = "_dollo_gl_secondary_features";
 147     public static final String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES                    = "_dollo_glc_secondary_features";
 148     public static final String                                PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES                               = "_dollo_gains_secondary_features";
 149     public static final String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES                              = "_dollo_losses_secondary_features";
 150     public static final String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES                             = "_dollo_present_secondary_features";
 151     public static final String                                SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                         = "_secondary_features_dollo"
 152                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 153     public static final String                                PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_ALL_NAMESPACES                              = "_dollo_goid_d";
 154     public static final String                                PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_ALL_NAMESPACES                             = "_fitch_goid_dc";
 155     final static private String                               HELP_OPTION_1                                                                 = "help";
 156     final static private String                               HELP_OPTION_2                                                                 = "h";
 157     final static private String                               OUTPUT_DIR_OPTION                                                             = "out_dir";
 158     final static private String                               SCORING_OPTION                                                                = "scoring";
 159     private static final DomainSimilarityScoring              SCORING_DEFAULT                                                               = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 160     final static private String                               SCORING_DOMAIN_COUNT_BASED                                                    = "domains";
 161     final static private String                               SCORING_PROTEIN_COUNT_BASED                                                   = "proteins";
 162     final static private String                               SCORING_COMBINATION_BASED                                                     = "combinations";
 163     final static private String                               DETAILEDNESS_OPTION                                                           = "detail";
 164     private final static Detailedness                         DETAILEDNESS_DEFAULT                                                          = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 165     final static private String                               SPECIES_MATRIX_OPTION                                                         = "smatrix";
 166     final static private String                               DETAILEDNESS_BASIC                                                            = "basic";
 167     final static private String                               DETAILEDNESS_LIST_IDS                                                         = "list_ids";
 168     final static private String                               DETAILEDNESS_PUNCTILIOUS                                                      = "punctilious";
 169     final static private String                               DOMAIN_SIMILARITY_SORT_OPTION                                                 = "sort";
 170     private static final DomainSimilaritySortField            DOMAIN_SORT_FILD_DEFAULT                                                      = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 171     final static private String                               DOMAIN_SIMILARITY_SORT_MIN                                                    = "min";
 172     final static private String                               DOMAIN_SIMILARITY_SORT_MAX                                                    = "max";
 173     final static private String                               DOMAIN_SIMILARITY_SORT_SD                                                     = "sd";
 174     final static private String                               DOMAIN_SIMILARITY_SORT_MEAN                                                   = "mean";
 175     final static private String                               DOMAIN_SIMILARITY_SORT_DIFF                                                   = "diff";
 176     final static private String                               DOMAIN_SIMILARITY_SORT_COUNTS_DIFF                                            = "count_diff";
 177     final static private String                               DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF                                        = "abs_count_diff";
 178     final static private String                               DOMAIN_SIMILARITY_SORT_SPECIES_COUNT                                          = "species";
 179     final static private String                               DOMAIN_SIMILARITY_SORT_ALPHA                                                  = "alpha";
 180     final static private String                               DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION                          = "species_first";
 181     final static private String                               DOMAIN_COUNT_SORT_OPTION                                                      = "dc_sort";
 182     private static final GenomeWideCombinableDomainsSortOrder DOMAINS_SORT_ORDER_DEFAULT                                                    = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
 183     final static private String                               DOMAIN_COUNT_SORT_ALPHA                                                       = "alpha";
 184     final static private String                               DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT                                            = "dom";
 185     final static private String                               DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT                                   = "prot";
 186     final static private String                               DOMAIN_COUNT_SORT_COMBINATIONS_COUNT                                          = "comb";
 187     final static private String                               CUTOFF_SCORE_FILE_OPTION                                                      = "cos";
 188     final static private String                               NOT_IGNORE_DUFS_OPTION                                                        = "dufs";
 189     final static private String                               MAX_E_VALUE_OPTION                                                            = "e";
 190     final static private String                               MAX_ALLOWED_OVERLAP_OPTION                                                    = "mo";
 191     final static private String                               NO_ENGULFING_OVERLAP_OPTION                                                   = "no_eo";
 192     final static private String                               IGNORE_COMBINATION_WITH_SAME_OPTION                                           = "ignore_self_comb";
 193     final static private String                               PAIRWISE_DOMAIN_COMPARISONS_PREFIX                                            = "pwc_";
 194     final static private String                               PAIRWISE_DOMAIN_COMPARISONS_OPTION                                            = "pwc";
 195     final static private String                               OUTPUT_FILE_OPTION                                                            = "o";
 196     final static private String                               PFAM_TO_GO_FILE_USE_OPTION                                                    = "p2g";
 197     final static private String                               GO_OBO_FILE_USE_OPTION                                                        = "obo";
 198     final static private String                               GO_NAMESPACE_LIMIT_OPTION                                                     = "go_namespace";
 199     final static private String                               GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION                                  = "molecular_function";
 200     final static private String                               GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS                                  = "biological_process";
 201     final static private String                               GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT                                  = "cellular_component";
 202     final static private String                               SECONDARY_FEATURES_PARSIMONY_MAP_FILE                                         = "secondary";
 203     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED                           = "simple_tab";
 204     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML                                    = "simple_html";
 205     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML                                  = "detailed_html";
 206     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION                                                = "ds_output";
 207     private static final PRINT_OPTION                         DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT                                        = PrintableDomainSimilarity.PRINT_OPTION.HTML;
 208     final static private String                               IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION                     = "ignore_singlet_domains";
 209     final static private String                               IGNORE_VIRAL_IDS                                                              = "ignore_viral_ids";
 210     final static private boolean                              IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT                    = false;
 211     final static private String                               IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION                                 = "ignore_species_specific_domains";
 212     final static private boolean                              IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT                         = false;
 213     final static private String                               MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                                = "_mean_score.pwd";
 214     final static private String                               MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                            = "_domains.pwd";
 215     final static private String                               MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX                   = "_bin_combinations.pwd";
 216     final static private String                               NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                               = "_mean_score_NJ"
 217                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 218     final static private String                               NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                           = "_domains_NJ"
 219                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 220     final static private String                               NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX                  = "_bin_combinations_NJ"
 221                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 222     final static private String                               JACKNIFE_OPTION                                                               = "jack";
 223     final static private String                               JACKNIFE_RANDOM_SEED_OPTION                                                   = "seed";
 224     final static private String                               JACKNIFE_RATIO_OPTION                                                         = "jack_ratio";
 225     private static final int                                  JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT                                        = 100;
 226     final static private long                                 JACKNIFE_RANDOM_SEED_DEFAULT                                                  = 19;
 227     final static private double                               JACKNIFE_RATIO_DEFAULT                                                        = 0.5;
 228     //final static private String  INFER_SPECIES_TREES_OPTION                                             = "species_tree_inference";
 229     final static private String                               INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX                                      = "_sd_nj.nh";
 230     final static private String                               INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX                                     = "_sbc_nj.nh";
 231     final static private String                               FILTER_POSITIVE_OPTION                                                        = "pos_filter";
 232     final static private String                               FILTER_NEGATIVE_OPTION                                                        = "neg_filter";
 233     final static private String                               FILTER_NEGATIVE_DOMAINS_OPTION                                                = "neg_dom_filter";
 234     final static private String                               INPUT_FILES_FROM_FILE_OPTION                                                  = "input";
 235     final static private String                               INPUT_SPECIES_TREE_OPTION                                                     = "species_tree";
 236     final static private String                               SEQ_EXTRACT_OPTION                                                            = "prot_extract";
 237     final static private char                                 SEPARATOR_FOR_INPUT_VALUES                                                    = '#';
 238     final static private String                               PRG_VERSION                                                                   = "2.250";
 239     final static private String                               PRG_DATE                                                                      = "2012.05.07";
 240     final static private String                               E_MAIL                                                                        = "czmasek@burnham.org";
 241     final static private String                               WWW                                                                           = "www.phylosoft.org/forester/applications/surfacing";
 242     final static private boolean                              IGNORE_DUFS_DEFAULT                                                           = true;
 243     final static private boolean                              IGNORE_COMBINATION_WITH_SAME_DEFAULLT                                         = false;
 244     final static private double                               MAX_E_VALUE_DEFAULT                                                           = -1;
 245     final static private int                                  MAX_ALLOWED_OVERLAP_DEFAULT                                                   = -1;
 246     private static final String                               RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION                                        = "random_seed";
 247     private static final String                               CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS                                      = "consider_bdc_direction";
 248     private static final String                               CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY                        = "consider_bdc_adj";
 249     private static final String                               SEQ_EXTRACT_SUFFIX                                                            = ".prot";
 250     private static final String                               PLUS_MINUS_ANALYSIS_OPTION                                                    = "plus_minus";
 251     private static final String                               PLUS_MINUS_DOM_SUFFIX                                                         = "_plus_minus_dom.txt";
 252     private static final String                               PLUS_MINUS_DOM_SUFFIX_HTML                                                    = "_plus_minus_dom.html";
 253     private static final String                               PLUS_MINUS_DC_SUFFIX_HTML                                                     = "_plus_minus_dc.html";
 254     private static final int                                  PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT                                          = 0;
 255     private static final double                               PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT                                            = 1.0;
 256     private static final String                               PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX                                              = "_plus_minus_go_ids_all.txt";
 257     private static final String                               PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX                                          = "_plus_minus_go_ids_passing.txt";
 258     private static final String                               OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS                                           = "all_prot";
 259     private static final boolean                              VERBOSE                                                                       = false;
 260     private static final String                               OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX              = "_fitch_dc_gains_counts";
 261     private static final String                               OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX                = "_fitch_dc_losses_counts";
 262     private static final String                               DOMAIN_LENGTHS_ANALYSIS_SUFFIX                                                = "_domain_lengths_analysis";
 263     private static final boolean                              PERFORM_DOMAIN_LENGTH_ANALYSIS                                                = true;
 264     public static final String                                ALL_PFAMS_ENCOUNTERED_SUFFIX                                                  = "_all_encountered_pfams";
 265     public static final String                                ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX                               = "_all_encountered_pfams_with_go_annotation";
 266     public static final String                                ENCOUNTERED_PFAMS_SUMMARY_SUFFIX                                              = "_encountered_pfams_summary";
 267     public static final String                                ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX                                            = "_all_pfams_gained_as_domains";
 268     public static final String                                ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX                                              = "_all_pfams_lost_as_domains";
 269     public static final String                                ALL_PFAMS_GAINED_AS_DC_SUFFIX                                                 = "_all_pfams_gained_as_dc";
 270     public static final String                                ALL_PFAMS_LOST_AS_DC_SUFFIX                                                   = "_all_pfams_lost_as_dc";
 271     public static final String                                BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES                                = "PER_NODE_EVENTS";
 272     public static final String                                BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES                             = "PER_SUBTREE_EVENTS";
 273     public static final String                                D_PROMISCUITY_FILE_SUFFIX                                                     = "_domain_promiscuities";
 274     private static final String                               LOG_FILE_SUFFIX                                                               = "_log.txt";
 275     private static final String                               DATA_FILE_SUFFIX                                                              = "_domain_combination_data.txt";
 276     private static final String                               DATA_FILE_DESC                                                                = "#SPECIES\tPRTEIN_ID\tN_TERM_DOMAIN\tC_TERM_DOMAIN\tN_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tC_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tN_TERM_DOMAIN_COUNTS_PER_PROTEIN\tC_TERM_DOMAIN_COUNTS_PER_PROTEIN";
 277     private static final INDIVIDUAL_SCORE_CUTOFF              INDIVIDUAL_SCORE_CUTOFF_DEFAULT                                               = INDIVIDUAL_SCORE_CUTOFF.FULL_SEQUENCE;
 278     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX                          = "_indep_dc_gains_fitch_counts.txt";
 279     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX                              = "_indep_dc_gains_fitch_lists.txt";
 280     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX               = "_indep_dc_gains_fitch_lists_for_go_mapping.txt";
 281     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX        = "_indep_dc_gains_fitch_lists_for_go_mapping_unique.txt";
 282     public static final String                                LIMIT_SPEC_FOR_PROT_EX                                                        = null;                                                                                                                                                                                       // e.g. "HUMAN"; set to null for not using this feature (default).
 283     public static final String                                BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH_MAPPED          = "_dc_MAPPED_secondary_features_fitch"
 284                                                                                                                                                     + ForesterConstants.PHYLO_XML_SUFFIX;
 285     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_MAPPED_OUTPUT_SUFFIX                   = "_indep_dc_gains_fitch_counts_MAPPED.txt";
 286     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_MAPPED_OUTPUT_SUFFIX                       = "_indep_dc_gains_fitch_lists_MAPPED.txt";
 287     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_SUFFIX        = "_indep_dc_gains_fitch_lists_for_go_mapping_MAPPED.txt";
 288     public static final String                                INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_UNIQUE_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping_unique_MAPPED.txt";
 289     private static final boolean                              PERFORM_DC_REGAIN_PROTEINS_STATS                                              = true;
 290
 291     private static void checkWriteabilityForPairwiseComparisons( final PrintableDomainSimilarity.PRINT_OPTION domain_similarity_print_option,
 292                                                                  final String[][] input_file_properties,
 293                                                                  final String automated_pairwise_comparison_suffix,
 294                                                                  final File outdir ) {
 295         for( int i = 0; i < input_file_properties.length; ++i ) {
 296             for( int j = 0; j < i; ++j ) {
 297                 final String species_i = input_file_properties[ i ][ 1 ];
 298                 final String species_j = input_file_properties[ j ][ 1 ];
 299                 String pairwise_similarities_output_file_str = PAIRWISE_DOMAIN_COMPARISONS_PREFIX + species_i + "_"
 300                         + species_j + automated_pairwise_comparison_suffix;
 301                 switch ( domain_similarity_print_option ) {
 302                     case HTML:
 303                         if ( !pairwise_similarities_output_file_str.endsWith( ".html" ) ) {
 304                             pairwise_similarities_output_file_str += ".html";
 305                         }
 306                         break;
 307                 }
 308                 final String error = ForesterUtil
 309                         .isWritableFile( new File( outdir == null ? pairwise_similarities_output_file_str : outdir
 310                                 + ForesterUtil.FILE_SEPARATOR + pairwise_similarities_output_file_str ) );
 311                 if ( !ForesterUtil.isEmpty( error ) ) {
 312                     ForesterUtil.fatalError( surfacing.PRG_NAME, error );
 313                 }
 314             }
 315         }
 316     }
 317
 318     private static StringBuilder createParametersAsString( final boolean ignore_dufs,
 319                                                            final double e_value_max,
 320                                                            final int max_allowed_overlap,
 321                                                            final boolean no_engulfing_overlaps,
 322                                                            final File cutoff_scores_file,
 323                                                            final BinaryDomainCombination.DomainCombinationType dc_type ) {
 324         final StringBuilder parameters_sb = new StringBuilder();
 325         parameters_sb.append( "E-value: " + e_value_max );
 326         if ( cutoff_scores_file != null ) {
 327             parameters_sb.append( ", Cutoff-scores-file: " + cutoff_scores_file );
 328         }
 329         else {
 330             parameters_sb.append( ", Cutoff-scores-file: not-set" );
 331         }
 332         if ( max_allowed_overlap != surfacing.MAX_ALLOWED_OVERLAP_DEFAULT ) {
 333             parameters_sb.append( ", Max-overlap: " + max_allowed_overlap );
 334         }
 335         else {
 336             parameters_sb.append( ", Max-overlap: not-set" );
 337         }
 338         if ( no_engulfing_overlaps ) {
 339             parameters_sb.append( ", Engulfing-overlaps: not-allowed" );
 340         }
 341         else {
 342             parameters_sb.append( ", Engulfing-overlaps: allowed" );
 343         }
 344         if ( ignore_dufs ) {
 345             parameters_sb.append( ", Ignore-dufs: true" );
 346         }
 347         else {
 348             parameters_sb.append( ", Ignore-dufs: false" );
 349         }
 350         parameters_sb.append( ", DC type (if applicable): " + dc_type );
 351         return parameters_sb;
 352     }
 353
 354     /**
 355      * Warning: This side-effects 'all_bin_domain_combinations_encountered'!
 356      *
 357      *
 358      * @param output_file
 359      * @param all_bin_domain_combinations_changed
 360      * @param sum_of_all_domains_encountered
 361      * @param all_bin_domain_combinations_encountered
 362      * @param is_gains_analysis
 363      * @param protein_length_stats_by_dc
 364      * @throws IOException
 365      */
 366     private static void executeFitchGainsAnalysis( final File output_file,
 367                                                    final List<BinaryDomainCombination> all_bin_domain_combinations_changed,
 368                                                    final int sum_of_all_domains_encountered,
 369                                                    final SortedSet<BinaryDomainCombination> all_bin_domain_combinations_encountered,
 370                                                    final boolean is_gains_analysis ) throws IOException {
 371         SurfacingUtil.checkForOutputFileWriteability( output_file );
 372         final Writer out = ForesterUtil.createBufferedWriter( output_file );
 373         final SortedMap<Object, Integer> bdc_to_counts = ForesterUtil
 374                 .listToSortedCountsMap( all_bin_domain_combinations_changed );
 375         final SortedSet<DomainId> all_domains_in_combination_changed_more_than_once = new TreeSet<DomainId>();
 376         final SortedSet<DomainId> all_domains_in_combination_changed_only_once = new TreeSet<DomainId>();
 377         int above_one = 0;
 378         int one = 0;
 379         for( final Object bdc_object : bdc_to_counts.keySet() ) {
 380             final BinaryDomainCombination bdc = ( BinaryDomainCombination ) bdc_object;
 381             final int count = bdc_to_counts.get( bdc_object );
 382             if ( count < 1 ) {
 383                 ForesterUtil.unexpectedFatalError( PRG_NAME, "count < 1 " );
 384             }
 385             out.write( bdc + "\t" + count + ForesterUtil.LINE_SEPARATOR );
 386             if ( count > 1 ) {
 387                 all_domains_in_combination_changed_more_than_once.add( bdc.getId0() );
 388                 all_domains_in_combination_changed_more_than_once.add( bdc.getId1() );
 389                 above_one++;
 390             }
 391             else if ( count == 1 ) {
 392                 all_domains_in_combination_changed_only_once.add( bdc.getId0() );
 393                 all_domains_in_combination_changed_only_once.add( bdc.getId1() );
 394                 one++;
 395             }
 396         }
 397         final int all = all_bin_domain_combinations_encountered.size();
 398         int never_lost = -1;
 399         if ( !is_gains_analysis ) {
 400             all_bin_domain_combinations_encountered.removeAll( all_bin_domain_combinations_changed );
 401             never_lost = all_bin_domain_combinations_encountered.size();
 402             for( final BinaryDomainCombination bdc : all_bin_domain_combinations_encountered ) {
 403                 out.write( bdc + "\t" + "0" + ForesterUtil.LINE_SEPARATOR );
 404             }
 405         }
 406         if ( is_gains_analysis ) {
 407             out.write( "Sum of all distinct domain combinations appearing once               : " + one
 408                     + ForesterUtil.LINE_SEPARATOR );
 409             out.write( "Sum of all distinct domain combinations appearing more than once     : " + above_one
 410                     + ForesterUtil.LINE_SEPARATOR );
 411             out.write( "Sum of all distinct domains in combinations apppearing only once     : "
 412                     + all_domains_in_combination_changed_only_once.size() + ForesterUtil.LINE_SEPARATOR );
 413             out.write( "Sum of all distinct domains in combinations apppearing more than once: "
 414                     + all_domains_in_combination_changed_more_than_once.size() + ForesterUtil.LINE_SEPARATOR );
 415         }
 416         else {
 417             out.write( "Sum of all distinct domain combinations never lost                   : " + never_lost
 418                     + ForesterUtil.LINE_SEPARATOR );
 419             out.write( "Sum of all distinct domain combinations lost once                    : " + one
 420                     + ForesterUtil.LINE_SEPARATOR );
 421             out.write( "Sum of all distinct domain combinations lost more than once          : " + above_one
 422                     + ForesterUtil.LINE_SEPARATOR );
 423             out.write( "Sum of all distinct domains in combinations lost only once           : "
 424                     + all_domains_in_combination_changed_only_once.size() + ForesterUtil.LINE_SEPARATOR );
 425             out.write( "Sum of all distinct domains in combinations lost more than once: "
 426                     + all_domains_in_combination_changed_more_than_once.size() + ForesterUtil.LINE_SEPARATOR );
 427         }
 428         out.write( "All binary combinations                                              : " + all
 429                 + ForesterUtil.LINE_SEPARATOR );
 430         out.write( "All domains                                                          : "
 431                 + sum_of_all_domains_encountered );
 432         out.close();
 433         ForesterUtil.programMessage( surfacing.PRG_NAME,
 434                                      "Wrote fitch domain combination dynamics counts analysis to \"" + output_file
 435                                              + "\"" );
 436     }
 437
 438     private static void executePlusMinusAnalysis( final File output_file,
 439                                                   final List<String> plus_minus_analysis_high_copy_base,
 440                                                   final List<String> plus_minus_analysis_high_copy_target,
 441                                                   final List<String> plus_minus_analysis_low_copy,
 442                                                   final List<GenomeWideCombinableDomains> gwcd_list,
 443                                                   final SortedMap<Species, List<Protein>> protein_lists_per_species,
 444                                                   final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 445                                                   final Map<GoId, GoTerm> go_id_to_term_map,
 446                                                   final List<Object> plus_minus_analysis_numbers ) {
 447         final Set<String> all_spec = new HashSet<String>();
 448         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
 449             all_spec.add( gwcd.getSpecies().getSpeciesId() );
 450         }
 451         final File html_out_dom = new File( output_file + PLUS_MINUS_DOM_SUFFIX_HTML );
 452         final File plain_out_dom = new File( output_file + PLUS_MINUS_DOM_SUFFIX );
 453         final File html_out_dc = new File( output_file + PLUS_MINUS_DC_SUFFIX_HTML );
 454         final File all_domains_go_ids_out_dom = new File( output_file + PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX );
 455         final File passing_domains_go_ids_out_dom = new File( output_file + PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX );
 456         final File proteins_file_base = new File( output_file + "" );
 457         final int min_diff = ( ( Integer ) plus_minus_analysis_numbers.get( 0 ) ).intValue();
 458         final double factor = ( ( Double ) plus_minus_analysis_numbers.get( 1 ) ).doubleValue();
 459         try {
 460             DomainCountsDifferenceUtil.calculateCopyNumberDifferences( gwcd_list,
 461                                                                        protein_lists_per_species,
 462                                                                        plus_minus_analysis_high_copy_base,
 463                                                                        plus_minus_analysis_high_copy_target,
 464                                                                        plus_minus_analysis_low_copy,
 465                                                                        min_diff,
 466                                                                        factor,
 467                                                                        plain_out_dom,
 468                                                                        html_out_dom,
 469                                                                        html_out_dc,
 470                                                                        domain_id_to_go_ids_map,
 471                                                                        go_id_to_term_map,
 472                                                                        all_domains_go_ids_out_dom,
 473                                                                        passing_domains_go_ids_out_dom,
 474                                                                        proteins_file_base );
 475         }
 476         catch ( final IOException e ) {
 477             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
 478         }
 479         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote plus minus domain analysis results to \""
 480                 + html_out_dom + "\"" );
 481         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote plus minus domain analysis results to \""
 482                 + plain_out_dom + "\"" );
 483         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote plus minus domain analysis results to \"" + html_out_dc
 484                 + "\"" );
 485         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote plus minus domain analysis based passing GO ids to \""
 486                 + passing_domains_go_ids_out_dom + "\"" );
 487         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote plus minus domain analysis based all GO ids to \""
 488                 + all_domains_go_ids_out_dom + "\"" );
 489     }
 490
 491     private static Phylogeny[] getIntrees( final File[] intree_files,
 492                                            final int number_of_genomes,
 493                                            final String[][] input_file_properties ) {
 494         final Phylogeny[] intrees = new Phylogeny[ intree_files.length ];
 495         int i = 0;
 496         for( final File intree_file : intree_files ) {
 497             Phylogeny intree = null;
 498             final String error = ForesterUtil.isReadableFile( intree_file );
 499             if ( !ForesterUtil.isEmpty( error ) ) {
 500                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read input tree file [" + intree_file + "]: "
 501                         + error );
 502             }
 503             try {
 504                 final Phylogeny[] p_array = ParserBasedPhylogenyFactory.getInstance()
 505                         .create( intree_file, ParserUtils.createParserDependingOnFileType( intree_file, true ) );
 506                 if ( p_array.length < 1 ) {
 507                     ForesterUtil.fatalError( surfacing.PRG_NAME, "file [" + intree_file
 508                             + "] does not contain any phylogeny in phyloXML format" );
 509                 }
 510                 else if ( p_array.length > 1 ) {
 511                     ForesterUtil.fatalError( surfacing.PRG_NAME, "file [" + intree_file
 512                             + "] contains more than one phylogeny in phyloXML format" );
 513                 }
 514                 intree = p_array[ 0 ];
 515             }
 516             catch ( final Exception e ) {
 517                 ForesterUtil.fatalError( surfacing.PRG_NAME, "failed to read input tree from file [" + intree_file
 518                         + "]: " + error );
 519             }
 520             if ( ( intree == null ) || intree.isEmpty() ) {
 521                 ForesterUtil.fatalError( surfacing.PRG_NAME, "input tree [" + intree_file + "] is empty" );
 522             }
 523             if ( !intree.isRooted() ) {
 524                 ForesterUtil.fatalError( surfacing.PRG_NAME, "input tree [" + intree_file + "] is not rooted" );
 525             }
 526             if ( intree.getNumberOfExternalNodes() < number_of_genomes ) {
 527                 ForesterUtil.fatalError( surfacing.PRG_NAME,
 528                                          "number of external nodes [" + intree.getNumberOfExternalNodes()
 529                                                  + "] of input tree [" + intree_file
 530                                                  + "] is smaller than the number of genomes the be analyzed ["
 531                                                  + number_of_genomes + "]" );
 532             }
 533             final StringBuilder parent_names = new StringBuilder();
 534             final int nodes_lacking_name = SurfacingUtil.getNumberOfNodesLackingName( intree, parent_names );
 535             if ( nodes_lacking_name > 0 ) {
 536                 ForesterUtil.fatalError( surfacing.PRG_NAME, "input tree [" + intree_file + "] has "
 537                         + nodes_lacking_name + " node(s) lacking a name [parent names:" + parent_names + "]" );
 538             }
 539             preparePhylogenyForParsimonyAnalyses( intree, input_file_properties );
 540             if ( !intree.isCompletelyBinary() ) {
 541                 ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "input tree [" + intree_file
 542                         + "] is not completely binary" );
 543             }
 544             intrees[ i++ ] = intree;
 545         }
 546         return intrees;
 547     }
 548
 549     private static List<Phylogeny> inferSpeciesTrees( final File outfile,
 550                                                       final List<BasicSymmetricalDistanceMatrix> distances_list ) {
 551         final NeighborJoining nj = NeighborJoining.createInstance();
 552         final List<Phylogeny> phylogenies = nj.execute( distances_list );
 553         final PhylogenyWriter w = new PhylogenyWriter();
 554         try {
 555             w.toNewHampshire( phylogenies, true, true, outfile, ";" );
 556         }
 557         catch ( final IOException e ) {
 558             ForesterUtil.fatalError( PRG_NAME, "failed to write to outfile [" + outfile + "]: " + e.getMessage() );
 559         }
 560         return phylogenies;
 561     }
 562
 563     private static void log( final String msg, final Writer w ) {
 564         try {
 565             w.write( msg );
 566             w.write( ForesterUtil.LINE_SEPARATOR );
 567         }
 568         catch ( final IOException e ) {
 569             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
 570         }
 571     }
 572
 573     public static void main( final String args[] ) {
 574         final long start_time = new Date().getTime();
 575         // final StringBuffer log = new StringBuffer();
 576         final StringBuilder html_desc = new StringBuilder();
 577         ForesterUtil.printProgramInformation( surfacing.PRG_NAME,
 578                                               surfacing.PRG_VERSION,
 579                                               surfacing.PRG_DATE,
 580                                               surfacing.E_MAIL,
 581                                               surfacing.WWW );
 582         final String nl = ForesterUtil.LINE_SEPARATOR;
 583         html_desc.append( "<table>" + nl );
 584         html_desc.append( "<tr><td>Produced by:</td><td>" + surfacing.PRG_NAME + "</td></tr>" + nl );
 585         html_desc.append( "<tr><td>Version:</td><td>" + surfacing.PRG_VERSION + "</td></tr>" + nl );
 586         html_desc.append( "<tr><td>Release Date:</td><td>" + surfacing.PRG_DATE + "</td></tr>" + nl );
 587         html_desc.append( "<tr><td>Contact:</td><td>" + surfacing.E_MAIL + "</td></tr>" + nl );
 588         html_desc.append( "<tr><td>WWW:</td><td>" + surfacing.WWW + "</td></tr>" + nl );
 589         CommandLineArguments cla = null;
 590         try {
 591             cla = new CommandLineArguments( args );
 592         }
 593         catch ( final Exception e ) {
 594             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
 595         }
 596         if ( cla.isOptionSet( surfacing.HELP_OPTION_1 ) || cla.isOptionSet( surfacing.HELP_OPTION_2 ) ) {
 597             surfacing.printHelp();
 598             System.exit( 0 );
 599         }
 600         if ( ( args.length < 1 ) ) {
 601             surfacing.printHelp();
 602             System.exit( -1 );
 603         }
 604         final List<String> allowed_options = new ArrayList<String>();
 605         allowed_options.add( surfacing.NOT_IGNORE_DUFS_OPTION );
 606         allowed_options.add( surfacing.MAX_E_VALUE_OPTION );
 607         allowed_options.add( surfacing.DETAILEDNESS_OPTION );
 608         allowed_options.add( surfacing.OUTPUT_FILE_OPTION );
 609         allowed_options.add( surfacing.DOMAIN_SIMILARITY_SORT_OPTION );
 610         allowed_options.add( surfacing.SPECIES_MATRIX_OPTION );
 611         allowed_options.add( surfacing.SCORING_OPTION );
 612         allowed_options.add( surfacing.MAX_ALLOWED_OVERLAP_OPTION );
 613         allowed_options.add( surfacing.NO_ENGULFING_OVERLAP_OPTION );
 614         allowed_options.add( surfacing.DOMAIN_COUNT_SORT_OPTION );
 615         allowed_options.add( surfacing.CUTOFF_SCORE_FILE_OPTION );
 616         allowed_options.add( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION );
 617         allowed_options.add( surfacing.OUTPUT_DIR_OPTION );
 618         allowed_options.add( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION );
 619         allowed_options.add( surfacing.PFAM_TO_GO_FILE_USE_OPTION );
 620         allowed_options.add( surfacing.GO_OBO_FILE_USE_OPTION );
 621         allowed_options.add( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION );
 622         allowed_options.add( surfacing.GO_NAMESPACE_LIMIT_OPTION );
 623         allowed_options.add( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 624         allowed_options.add( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION );
 625         allowed_options.add( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS );
 626         allowed_options.add( JACKNIFE_OPTION );
 627         allowed_options.add( JACKNIFE_RANDOM_SEED_OPTION );
 628         allowed_options.add( JACKNIFE_RATIO_OPTION );
 629         allowed_options.add( INPUT_SPECIES_TREE_OPTION );
 630         //allowed_options.add( INFER_SPECIES_TREES_OPTION );
 631         allowed_options.add( FILTER_POSITIVE_OPTION );
 632         allowed_options.add( FILTER_NEGATIVE_OPTION );
 633         allowed_options.add( INPUT_FILES_FROM_FILE_OPTION );
 634         allowed_options.add( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
 635         allowed_options.add( FILTER_NEGATIVE_DOMAINS_OPTION );
 636         allowed_options.add( IGNORE_VIRAL_IDS );
 637         allowed_options.add( SEQ_EXTRACT_OPTION );
 638         allowed_options.add( SECONDARY_FEATURES_PARSIMONY_MAP_FILE );
 639         allowed_options.add( PLUS_MINUS_ANALYSIS_OPTION );
 640         allowed_options.add( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS );
 641         allowed_options.add( OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS );
 642         allowed_options.add( CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY );
 643         boolean ignore_dufs = surfacing.IGNORE_DUFS_DEFAULT;
 644         boolean ignore_combination_with_same = surfacing.IGNORE_COMBINATION_WITH_SAME_DEFAULLT;
 645         double e_value_max = surfacing.MAX_E_VALUE_DEFAULT;
 646         int max_allowed_overlap = surfacing.MAX_ALLOWED_OVERLAP_DEFAULT;
 647         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 648         if ( dissallowed_options.length() > 0 ) {
 649             ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown option(s): " + dissallowed_options );
 650         }
 651         boolean output_binary_domain_combinationsfor_graph_analysis = false;
 652         if ( cla.isOptionSet( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS ) ) {
 653             output_binary_domain_combinationsfor_graph_analysis = true;
 654         }
 655         if ( cla.isOptionSet( surfacing.MAX_E_VALUE_OPTION ) ) {
 656             try {
 657                 e_value_max = cla.getOptionValueAsDouble( surfacing.MAX_E_VALUE_OPTION );
 658             }
 659             catch ( final Exception e ) {
 660                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for E-value maximum" );
 661             }
 662         }
 663         if ( cla.isOptionSet( surfacing.MAX_ALLOWED_OVERLAP_OPTION ) ) {
 664             try {
 665                 max_allowed_overlap = cla.getOptionValueAsInt( surfacing.MAX_ALLOWED_OVERLAP_OPTION );
 666             }
 667             catch ( final Exception e ) {
 668                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for maximal allowed domain overlap" );
 669             }
 670         }
 671         boolean no_engulfing_overlaps = false;
 672         if ( cla.isOptionSet( surfacing.NO_ENGULFING_OVERLAP_OPTION ) ) {
 673             no_engulfing_overlaps = true;
 674         }
 675         boolean ignore_virus_like_ids = false;
 676         if ( cla.isOptionSet( surfacing.IGNORE_VIRAL_IDS ) ) {
 677             ignore_virus_like_ids = true;
 678         }
 679         if ( cla.isOptionSet( surfacing.NOT_IGNORE_DUFS_OPTION ) ) {
 680             ignore_dufs = false;
 681         }
 682         if ( cla.isOptionSet( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION ) ) {
 683             ignore_combination_with_same = true;
 684         }
 685         boolean ignore_domains_without_combs_in_all_spec = IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT;
 686         if ( cla.isOptionSet( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION ) ) {
 687             ignore_domains_without_combs_in_all_spec = true;
 688         }
 689         boolean ignore_species_specific_domains = IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT;
 690         if ( cla.isOptionSet( surfacing.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION ) ) {
 691             ignore_species_specific_domains = true;
 692         }
 693         File output_file = null;
 694         if ( cla.isOptionSet( surfacing.OUTPUT_FILE_OPTION ) ) {
 695             if ( !cla.isOptionValueSet( surfacing.OUTPUT_FILE_OPTION ) ) {
 696                 ForesterUtil.fatalError( surfacing.PRG_NAME,
 697                                          "no value for domain combinations similarities output file: -"
 698                                                  + surfacing.OUTPUT_FILE_OPTION + "=<file>" );
 699             }
 700             output_file = new File( cla.getOptionValue( surfacing.OUTPUT_FILE_OPTION ) );
 701             SurfacingUtil.checkForOutputFileWriteability( output_file );
 702         }
 703         File cutoff_scores_file = null;
 704         Map<String, Double> individual_score_cutoffs = null;
 705         if ( cla.isOptionSet( surfacing.CUTOFF_SCORE_FILE_OPTION ) ) {
 706             if ( !cla.isOptionValueSet( surfacing.CUTOFF_SCORE_FILE_OPTION ) ) {
 707                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for individual domain score cutoffs file: -"
 708                         + surfacing.CUTOFF_SCORE_FILE_OPTION + "=<file>" );
 709             }
 710             cutoff_scores_file = new File( cla.getOptionValue( surfacing.CUTOFF_SCORE_FILE_OPTION ) );
 711             final String error = ForesterUtil.isReadableFile( cutoff_scores_file );
 712             if ( !ForesterUtil.isEmpty( error ) ) {
 713                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read individual domain score cutoffs file: "
 714                         + error );
 715             }
 716             try {
 717                 final BasicTable<String> scores_table = BasicTableParser.parse( cutoff_scores_file, " " );
 718                 individual_score_cutoffs = scores_table.getColumnsAsMapDouble( 0, 1 );
 719             }
 720             catch ( final IOException e ) {
 721                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from individual score cutoffs file: " + e );
 722             }
 723         }
 724         BinaryDomainCombination.DomainCombinationType dc_type = BinaryDomainCombination.DomainCombinationType.BASIC;
 725         if ( cla.isOptionSet( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS ) ) {
 726             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED;
 727         }
 728         if ( cla.isOptionSet( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY ) ) {
 729             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT;
 730         }
 731         File out_dir = null;
 732         if ( cla.isOptionSet( surfacing.OUTPUT_DIR_OPTION ) ) {
 733             if ( !cla.isOptionValueSet( surfacing.OUTPUT_DIR_OPTION ) ) {
 734                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for output directory: -"
 735                         + surfacing.OUTPUT_DIR_OPTION + "=<dir>" );
 736             }
 737             out_dir = new File( cla.getOptionValue( surfacing.OUTPUT_DIR_OPTION ) );
 738             if ( out_dir.exists() && ( out_dir.listFiles().length > 0 ) ) {
 739                 ForesterUtil.fatalError( surfacing.PRG_NAME, "\"" + out_dir + "\" aready exists and is not empty" );
 740             }
 741             if ( !out_dir.exists() ) {
 742                 final boolean success = out_dir.mkdir();
 743                 if ( !success || !out_dir.exists() ) {
 744                     ForesterUtil.fatalError( surfacing.PRG_NAME, "failed to create \"" + out_dir + "\"" );
 745                 }
 746             }
 747             if ( !out_dir.canWrite() ) {
 748                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot write to \"" + out_dir + "\"" );
 749             }
 750         }
 751         File positive_filter_file = null;
 752         File negative_filter_file = null;
 753         File negative_domains_filter_file = null;
 754         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) && cla.isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 755             ForesterUtil.fatalError( surfacing.PRG_NAME, "attempt to use both negative and positive protein filter" );
 756         }
 757         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION )
 758                 && ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) || cla
 759                         .isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) ) {
 760             ForesterUtil
 761                     .fatalError( surfacing.PRG_NAME,
 762                                  "attempt to use both negative or positive protein filter together wirh a negative domains filter" );
 763         }
 764         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) ) {
 765             if ( !cla.isOptionValueSet( surfacing.FILTER_NEGATIVE_OPTION ) ) {
 766                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for negative filter: -"
 767                         + surfacing.FILTER_NEGATIVE_OPTION + "=<file>" );
 768             }
 769             negative_filter_file = new File( cla.getOptionValue( surfacing.FILTER_NEGATIVE_OPTION ) );
 770             final String msg = ForesterUtil.isReadableFile( negative_filter_file );
 771             if ( !ForesterUtil.isEmpty( msg ) ) {
 772                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + negative_filter_file + "\": "
 773                         + msg );
 774             }
 775         }
 776         else if ( cla.isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 777             if ( !cla.isOptionValueSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 778                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for positive filter: -"
 779                         + surfacing.FILTER_POSITIVE_OPTION + "=<file>" );
 780             }
 781             positive_filter_file = new File( cla.getOptionValue( surfacing.FILTER_POSITIVE_OPTION ) );
 782             final String msg = ForesterUtil.isReadableFile( positive_filter_file );
 783             if ( !ForesterUtil.isEmpty( msg ) ) {
 784                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + positive_filter_file + "\": "
 785                         + msg );
 786             }
 787         }
 788         else if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 789             if ( !cla.isOptionValueSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 790                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for negative domains filter: -"
 791                         + surfacing.FILTER_NEGATIVE_DOMAINS_OPTION + "=<file>" );
 792             }
 793             negative_domains_filter_file = new File( cla.getOptionValue( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) );
 794             final String msg = ForesterUtil.isReadableFile( negative_domains_filter_file );
 795             if ( !ForesterUtil.isEmpty( msg ) ) {
 796                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + negative_domains_filter_file
 797                         + "\": " + msg );
 798             }
 799         }
 800         final List<String> plus_minus_analysis_high_copy_base_species = new ArrayList<String>();
 801         final List<String> plus_minus_analysis_high_copy_target_species = new ArrayList<String>();
 802         final List<String> plus_minus_analysis_high_low_copy_species = new ArrayList<String>();
 803         final List<Object> plus_minus_analysis_numbers = new ArrayList<Object>();
 804         processPlusMinusAnalysisOption( cla,
 805                                         plus_minus_analysis_high_copy_base_species,
 806                                         plus_minus_analysis_high_copy_target_species,
 807                                         plus_minus_analysis_high_low_copy_species,
 808                                         plus_minus_analysis_numbers );
 809         File input_files_file = null;
 810         String[] input_file_names_from_file = null;
 811         if ( cla.isOptionSet( surfacing.INPUT_FILES_FROM_FILE_OPTION ) ) {
 812             if ( !cla.isOptionValueSet( surfacing.INPUT_FILES_FROM_FILE_OPTION ) ) {
 813                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for input files file: -"
 814                         + surfacing.INPUT_FILES_FROM_FILE_OPTION + "=<file>" );
 815             }
 816             input_files_file = new File( cla.getOptionValue( surfacing.INPUT_FILES_FROM_FILE_OPTION ) );
 817             final String msg = ForesterUtil.isReadableFile( input_files_file );
 818             if ( !ForesterUtil.isEmpty( msg ) ) {
 819                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + input_files_file + "\": " + msg );
 820             }
 821             try {
 822                 input_file_names_from_file = ForesterUtil.file2array( input_files_file );
 823             }
 824             catch ( final IOException e ) {
 825                 ForesterUtil.fatalError( surfacing.PRG_NAME, "failed to read from \"" + input_files_file + "\": " + e );
 826             }
 827         }
 828         if ( ( cla.getNumberOfNames() < 1 )
 829                 && ( ( input_file_names_from_file == null ) || ( input_file_names_from_file.length < 1 ) ) ) {
 830             ForesterUtil.fatalError( surfacing.PRG_NAME,
 831                                      "No hmmpfam output file indicated is input: use comand line directly or "
 832                                              + surfacing.INPUT_FILES_FROM_FILE_OPTION + "=<file>" );
 833         }
 834         DomainSimilarity.DomainSimilarityScoring scoring = SCORING_DEFAULT;
 835         if ( cla.isOptionSet( surfacing.SCORING_OPTION ) ) {
 836             if ( !cla.isOptionValueSet( surfacing.SCORING_OPTION ) ) {
 837                 ForesterUtil.fatalError( surfacing.PRG_NAME,
 838                                          "no value for scoring method for domain combinations similarity calculation: -"
 839                                                  + surfacing.SCORING_OPTION + "=<"
 840                                                  + surfacing.SCORING_DOMAIN_COUNT_BASED + "|"
 841                                                  + surfacing.SCORING_PROTEIN_COUNT_BASED + "|"
 842                                                  + surfacing.SCORING_COMBINATION_BASED + ">\"" );
 843             }
 844             final String scoring_str = cla.getOptionValue( surfacing.SCORING_OPTION );
 845             if ( scoring_str.equals( surfacing.SCORING_DOMAIN_COUNT_BASED ) ) {
 846                 scoring = DomainSimilarity.DomainSimilarityScoring.DOMAINS;
 847             }
 848             else if ( scoring_str.equals( surfacing.SCORING_COMBINATION_BASED ) ) {
 849                 scoring = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 850             }
 851             else if ( scoring_str.equals( surfacing.SCORING_PROTEIN_COUNT_BASED ) ) {
 852                 scoring = DomainSimilarity.DomainSimilarityScoring.PROTEINS;
 853             }
 854             else {
 855                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + scoring_str
 856                         + "\" for scoring method for domain combinations similarity calculation: \"-"
 857                         + surfacing.SCORING_OPTION + "=<" + surfacing.SCORING_DOMAIN_COUNT_BASED + "|"
 858                         + surfacing.SCORING_PROTEIN_COUNT_BASED + "|" + surfacing.SCORING_COMBINATION_BASED + ">\"" );
 859             }
 860         }
 861         boolean sort_by_species_count_first = false;
 862         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION ) ) {
 863             sort_by_species_count_first = true;
 864         }
 865         boolean species_matrix = false;
 866         if ( cla.isOptionSet( surfacing.SPECIES_MATRIX_OPTION ) ) {
 867             species_matrix = true;
 868         }
 869         boolean output_protein_lists_for_all_domains = false;
 870         if ( cla.isOptionSet( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS ) ) {
 871             output_protein_lists_for_all_domains = true;
 872         }
 873         Detailedness detailedness = DETAILEDNESS_DEFAULT;
 874         if ( cla.isOptionSet( surfacing.DETAILEDNESS_OPTION ) ) {
 875             if ( !cla.isOptionValueSet( surfacing.DETAILEDNESS_OPTION ) ) {
 876                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for -" + surfacing.DETAILEDNESS_OPTION + "=<"
 877                         + surfacing.DETAILEDNESS_BASIC + "|" + surfacing.DETAILEDNESS_LIST_IDS + "|"
 878                         + surfacing.DETAILEDNESS_PUNCTILIOUS + ">\"" );
 879             }
 880             final String detness = cla.getOptionValue( surfacing.DETAILEDNESS_OPTION ).toLowerCase();
 881             if ( detness.equals( surfacing.DETAILEDNESS_BASIC ) ) {
 882                 detailedness = DomainSimilarityCalculator.Detailedness.BASIC;
 883             }
 884             else if ( detness.equals( surfacing.DETAILEDNESS_LIST_IDS ) ) {
 885                 detailedness = DomainSimilarityCalculator.Detailedness.LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES;
 886             }
 887             else if ( detness.equals( surfacing.DETAILEDNESS_PUNCTILIOUS ) ) {
 888                 detailedness = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 889             }
 890             else {
 891                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + detness + "\" for detailedness: \"-"
 892                         + surfacing.DETAILEDNESS_OPTION + "=<" + surfacing.DETAILEDNESS_BASIC + "|"
 893                         + surfacing.DETAILEDNESS_LIST_IDS + "|" + surfacing.DETAILEDNESS_PUNCTILIOUS + ">\"" );
 894             }
 895         }
 896         String automated_pairwise_comparison_suffix = null;
 897         boolean perform_pwc = false;
 898         boolean write_pwc_files = false;
 899         if ( cla.isOptionSet( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 900             perform_pwc = true;
 901             if ( !cla.isOptionValueSet( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 902                 write_pwc_files = false;
 903             }
 904             else {
 905                 write_pwc_files = true;
 906                 automated_pairwise_comparison_suffix = "_"
 907                         + cla.getOptionValue( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 908             }
 909         }
 910         String query_domain_ids = null;
 911         if ( cla.isOptionSet( surfacing.SEQ_EXTRACT_OPTION ) ) {
 912             if ( !cla.isOptionValueSet( surfacing.SEQ_EXTRACT_OPTION ) ) {
 913                 ForesterUtil
 914                         .fatalError( surfacing.PRG_NAME,
 915                                      "no domain ids given for sequences with given domains to be extracted : -"
 916                                              + surfacing.SEQ_EXTRACT_OPTION
 917                                              + "=<ordered domain sequences, domain ids separated by '~', sequences separated by '#'>" );
 918             }
 919             query_domain_ids = cla.getOptionValue( surfacing.SEQ_EXTRACT_OPTION );
 920         }
 921         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field = DOMAIN_SORT_FILD_DEFAULT;
 922         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field_for_automated_pwc = DOMAIN_SORT_FILD_DEFAULT;
 923         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 924             if ( !cla.isOptionValueSet( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 925                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for domain combinations similarities sorting: -"
 926                         + surfacing.DOMAIN_SIMILARITY_SORT_OPTION + "=<" + surfacing.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 927                         + surfacing.DOMAIN_SIMILARITY_SORT_MAX + "|" + surfacing.DOMAIN_SIMILARITY_SORT_MIN + "|"
 928                         + surfacing.DOMAIN_SIMILARITY_SORT_MEAN + "|" + surfacing.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 929                         + surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 930                         + surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|"
 931                         + surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|" + surfacing.DOMAIN_SIMILARITY_SORT_SD
 932                         + ">\"" );
 933             }
 934             final String sort_str = cla.getOptionValue( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ).toLowerCase();
 935             if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_ALPHA ) ) {
 936                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 937                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 938             }
 939             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MAX ) ) {
 940                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX;
 941                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 942             }
 943             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MIN ) ) {
 944                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MIN;
 945                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 946             }
 947             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MEAN ) ) {
 948                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MEAN;
 949                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MEAN;
 950             }
 951             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT ) ) {
 952                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SPECIES_COUNT;
 953                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 954             }
 955             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_SD ) ) {
 956                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SD;
 957                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 958             }
 959             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_DIFF ) ) {
 960                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 961                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 962             }
 963             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF ) ) {
 964                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 965                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 966             }
 967             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF ) ) {
 968                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 969                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 970             }
 971             else {
 972                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort_str
 973                         + "\" for domain combinations similarities sorting: \"-"
 974                         + surfacing.DOMAIN_SIMILARITY_SORT_OPTION + "=<" + surfacing.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 975                         + surfacing.DOMAIN_SIMILARITY_SORT_MAX + "|" + surfacing.DOMAIN_SIMILARITY_SORT_MIN + "|"
 976                         + surfacing.DOMAIN_SIMILARITY_SORT_MEAN + "|" + surfacing.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 977                         + surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 978                         + surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|" + "|"
 979                         + surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|" + surfacing.DOMAIN_SIMILARITY_SORT_SD
 980                         + ">\"" );
 981             }
 982         }
 983         PrintableDomainSimilarity.PRINT_OPTION domain_similarity_print_option = DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT;
 984         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 985             if ( !cla.isOptionValueSet( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 986                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for print option: -"
 987                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML + "|"
 988                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
 989                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
 990             }
 991             final String sort = cla.getOptionValue( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ).toLowerCase();
 992             if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML ) ) {
 993                 domain_similarity_print_option = PrintableDomainSimilarity.PRINT_OPTION.HTML;
 994             }
 995             else if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML ) ) {
 996                 // domain_similarity_print_option =
 997                 // DomainSimilarity.PRINT_OPTION.SIMPLE_HTML;
 998                 ForesterUtil.fatalError( surfacing.PRG_NAME, "simple HTML output not implemented yet :(" );
 999             }
1000             else if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED ) ) {
1001                 domain_similarity_print_option = PrintableDomainSimilarity.PRINT_OPTION.SIMPLE_TAB_DELIMITED;
1002             }
1003             else {
1004                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort + "\" for print option: -"
1005                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML + "|"
1006                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
1007                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
1008             }
1009         }
1010         GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order = DOMAINS_SORT_ORDER_DEFAULT;
1011         if ( cla.isOptionSet( surfacing.DOMAIN_COUNT_SORT_OPTION ) ) {
1012             if ( !cla.isOptionValueSet( surfacing.DOMAIN_COUNT_SORT_OPTION ) ) {
1013                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for sorting of domain counts: -"
1014                         + surfacing.DOMAIN_COUNT_SORT_OPTION + "=<" + surfacing.DOMAIN_COUNT_SORT_ALPHA + "|"
1015                         + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
1016                         + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
1017                         + surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
1018             }
1019             final String sort = cla.getOptionValue( surfacing.DOMAIN_COUNT_SORT_OPTION ).toLowerCase();
1020             if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_ALPHA ) ) {
1021                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
1022             }
1023             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT ) ) {
1024                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_COUNT;
1025             }
1026             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT ) ) {
1027                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_PROTEINS_COUNT;
1028             }
1029             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT ) ) {
1030                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.COMBINATIONS_COUNT;
1031             }
1032             else {
1033                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort
1034                         + "\" for sorting of domain counts: \"-" + surfacing.DOMAIN_COUNT_SORT_OPTION + "=<"
1035                         + surfacing.DOMAIN_COUNT_SORT_ALPHA + "|" + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
1036                         + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
1037                         + surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
1038             }
1039         }
1040         String[][] input_file_properties = null;
1041         if ( input_file_names_from_file != null ) {
1042             input_file_properties = surfacing.processInputFileNames( input_file_names_from_file );
1043         }
1044         else {
1045             input_file_properties = surfacing.processInputFileNames( cla.getNames() );
1046         }
1047         final int number_of_genomes = input_file_properties.length;
1048         if ( number_of_genomes < 2 ) {
1049             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot analyze less than two files" );
1050         }
1051         if ( ( number_of_genomes < 3 ) && perform_pwc ) {
1052             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use : -"
1053                     + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1054                     + "=<suffix> to turn on pairwise analyses with less than three input files" );
1055         }
1056         checkWriteabilityForPairwiseComparisons( domain_similarity_print_option,
1057                                                  input_file_properties,
1058                                                  automated_pairwise_comparison_suffix,
1059                                                  out_dir );
1060         for( int i = 0; i < number_of_genomes; i++ ) {
1061             File dcc_outfile = new File( input_file_properties[ i ][ 0 ]
1062                     + surfacing.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
1063             if ( out_dir != null ) {
1064                 dcc_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + dcc_outfile );
1065             }
1066             SurfacingUtil.checkForOutputFileWriteability( dcc_outfile );
1067         }
1068         File pfam_to_go_file = null;
1069         Map<DomainId, List<GoId>> domain_id_to_go_ids_map = null;
1070         int domain_id_to_go_ids_count = 0;
1071         if ( cla.isOptionSet( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) ) {
1072             if ( !cla.isOptionValueSet( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) ) {
1073                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for Pfam to GO mapping file: -"
1074                         + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>" );
1075             }
1076             pfam_to_go_file = new File( cla.getOptionValue( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) );
1077             final String error = ForesterUtil.isReadableFile( pfam_to_go_file );
1078             if ( !ForesterUtil.isEmpty( error ) ) {
1079                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read Pfam to GO mapping file: " + error );
1080             }
1081             try {
1082                 final PfamToGoParser parser = new PfamToGoParser( pfam_to_go_file );
1083                 final List<PfamToGoMapping> pfam_to_go_mappings = parser.parse();
1084                 domain_id_to_go_ids_map = SurfacingUtil.createDomainIdToGoIdMap( pfam_to_go_mappings );
1085                 if ( parser.getMappingCount() < domain_id_to_go_ids_map.size() ) {
1086                     ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME,
1087                                                        "parser.getMappingCount() < domain_id_to_go_ids_map.size()" );
1088                 }
1089                 domain_id_to_go_ids_count = parser.getMappingCount();
1090             }
1091             catch ( final IOException e ) {
1092                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from Pfam to GO mapping file: " + e );
1093             }
1094         }
1095         File go_obo_file = null;
1096         List<GoTerm> go_terms = null;
1097         if ( cla.isOptionSet( surfacing.GO_OBO_FILE_USE_OPTION ) ) {
1098             if ( !cla.isOptionValueSet( surfacing.GO_OBO_FILE_USE_OPTION ) ) {
1099                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for GO OBO file: -"
1100                         + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>" );
1101             }
1102             if ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) {
1103                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use GO OBO file (-"
1104                         + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>) without Pfam to GO mapping file ("
1105                         + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>)" );
1106             }
1107             go_obo_file = new File( cla.getOptionValue( surfacing.GO_OBO_FILE_USE_OPTION ) );
1108             final String error = ForesterUtil.isReadableFile( go_obo_file );
1109             if ( !ForesterUtil.isEmpty( error ) ) {
1110                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read GO OBO file: " + error );
1111             }
1112             try {
1113                 final OBOparser parser = new OBOparser( go_obo_file, OBOparser.ReturnType.BASIC_GO_TERM );
1114                 go_terms = parser.parse();
1115                 if ( parser.getGoTermCount() != go_terms.size() ) {
1116                     ForesterUtil
1117                             .unexpectedFatalError( surfacing.PRG_NAME, "parser.getGoTermCount() != go_terms.size()" );
1118                 }
1119             }
1120             catch ( final IOException e ) {
1121                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from GO OBO file: " + e );
1122             }
1123         }
1124         Map<GoId, GoTerm> go_id_to_term_map = null;
1125         if ( ( ( domain_id_to_go_ids_map != null ) && ( domain_id_to_go_ids_map.size() > 0 ) )
1126                 && ( ( go_terms != null ) && ( go_terms.size() > 0 ) ) ) {
1127             go_id_to_term_map = GoUtils.createGoIdToGoTermMap( go_terms );
1128         }
1129         GoNameSpace go_namespace_limit = null;
1130         if ( cla.isOptionSet( surfacing.GO_NAMESPACE_LIMIT_OPTION ) ) {
1131             if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
1132                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use GO namespace limit (-"
1133                         + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<namespace>) without Pfam to GO mapping file ("
1134                         + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>) and GO OBO file (-"
1135                         + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>)" );
1136             }
1137             if ( !cla.isOptionValueSet( surfacing.GO_NAMESPACE_LIMIT_OPTION ) ) {
1138                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for GO namespace limit: \"-"
1139                         + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<"
1140                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
1141                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
1142                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
1143             }
1144             final String go_namespace_limit_str = cla.getOptionValue( surfacing.GO_NAMESPACE_LIMIT_OPTION )
1145                     .toLowerCase();
1146             if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION ) ) {
1147                 go_namespace_limit = GoNameSpace.createMolecularFunction();
1148             }
1149             else if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS ) ) {
1150                 go_namespace_limit = GoNameSpace.createBiologicalProcess();
1151             }
1152             else if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT ) ) {
1153                 go_namespace_limit = GoNameSpace.createCellularComponent();
1154             }
1155             else {
1156                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + go_namespace_limit_str
1157                         + "\" for GO namespace limit: \"-" + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<"
1158                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
1159                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
1160                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
1161             }
1162         }
1163         if ( ( domain_similarity_sort_field == DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE )
1164                 && ( number_of_genomes > 2 ) ) {
1165             domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
1166         }
1167         boolean jacknifed_distances = false;
1168         int jacknife_resamplings = JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT;
1169         double jacknife_ratio = JACKNIFE_RATIO_DEFAULT;
1170         long random_seed = JACKNIFE_RANDOM_SEED_DEFAULT;
1171         if ( cla.isOptionSet( surfacing.JACKNIFE_OPTION ) ) {
1172             if ( ( number_of_genomes < 3 ) || !perform_pwc ) {
1173                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use jacknife resampling analysis (-"
1174                         + surfacing.JACKNIFE_OPTION + "[=<number of resamplings>]) without pairwise analyses ("
1175                         + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1176                         + "=<suffix for pairwise comparison output files>)" );
1177             }
1178             jacknifed_distances = true;
1179             if ( cla.isOptionHasAValue( surfacing.JACKNIFE_OPTION ) ) {
1180                 try {
1181                     jacknife_resamplings = cla.getOptionValueAsInt( surfacing.JACKNIFE_OPTION );
1182                 }
1183                 catch ( final IOException e ) {
1184                     ForesterUtil.fatalError( surfacing.PRG_NAME, "illegal format for number of resamplings" );
1185                 }
1186                 if ( jacknife_resamplings < 2 ) {
1187                     ForesterUtil.fatalError( surfacing.PRG_NAME, "attempt to use less than 2 resamplings" );
1188                 }
1189             }
1190             if ( cla.isOptionSet( surfacing.JACKNIFE_RATIO_OPTION )
1191                     && cla.isOptionHasAValue( surfacing.JACKNIFE_RATIO_OPTION ) ) {
1192                 try {
1193                     jacknife_ratio = cla.getOptionValueAsDouble( surfacing.JACKNIFE_RATIO_OPTION );
1194                 }
1195                 catch ( final IOException e ) {
1196                     ForesterUtil.fatalError( surfacing.PRG_NAME, "illegal format for jacknife ratio" );
1197                 }
1198                 if ( ( jacknife_ratio <= 0.0 ) || ( jacknife_ratio >= 1.0 ) ) {
1199                     ForesterUtil.fatalError( surfacing.PRG_NAME, "attempt to use illegal value for jacknife ratio: "
1200                             + jacknife_ratio );
1201                 }
1202             }
1203             if ( cla.isOptionSet( surfacing.JACKNIFE_RANDOM_SEED_OPTION )
1204                     && cla.isOptionHasAValue( surfacing.JACKNIFE_RANDOM_SEED_OPTION ) ) {
1205                 try {
1206                     random_seed = cla.getOptionValueAsLong( surfacing.JACKNIFE_RANDOM_SEED_OPTION );
1207                 }
1208                 catch ( final IOException e ) {
1209                     ForesterUtil.fatalError( surfacing.PRG_NAME, "illegal format for random generator seed" );
1210                 }
1211             }
1212         }
1213         //        boolean infer_species_trees = false;
1214         //        if ( cla.isOptionSet( surfacing.INFER_SPECIES_TREES_OPTION ) ) {
1215         //            if ( ( output_file == null ) || ( number_of_genomes < 3 )
1216         //                    || ForesterUtil.isEmpty( automated_pairwise_comparison_suffix ) ) {
1217         //                ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot infer species trees (-"
1218         //                        + surfacing.INFER_SPECIES_TREES_OPTION + " without pairwise analyses ("
1219         //                        + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1220         //                        + "=<suffix for pairwise comparison output files>)" );
1221         //            }
1222         //            infer_species_trees = true;
1223         //        }
1224         File[] intree_files = null;
1225         Phylogeny[] intrees = null;
1226         if ( cla.isOptionSet( surfacing.INPUT_SPECIES_TREE_OPTION ) ) {
1227             // TODO FIXME if jacknife.... maybe not
1228             if ( number_of_genomes < 3 ) {
1229                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot infer gains and losses on input species trees (-"
1230                         + surfacing.INPUT_SPECIES_TREE_OPTION + " without pairwise analyses ("
1231                         + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1232                         + "=<suffix for pairwise comparison output files>)" );
1233             }
1234             if ( !cla.isOptionValueSet( surfacing.INPUT_SPECIES_TREE_OPTION ) ) {
1235                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for input tree: -"
1236                         + surfacing.INPUT_SPECIES_TREE_OPTION + "=<tree file in phyloXML format>" );
1237             }
1238             final String intrees_str = cla.getOptionValue( surfacing.INPUT_SPECIES_TREE_OPTION );
1239             if ( intrees_str.indexOf( "#" ) > 0 ) {
1240                 final String[] intrees_strs = intrees_str.split( "#" );
1241                 intree_files = new File[ intrees_strs.length ];
1242                 int i = 0;
1243                 for( final String s : intrees_strs ) {
1244                     intree_files[ i++ ] = new File( s.trim() );
1245                 }
1246             }
1247             else {
1248                 intree_files = new File[ 1 ];
1249                 intree_files[ 0 ] = new File( intrees_str );
1250             }
1251             intrees = getIntrees( intree_files, number_of_genomes, input_file_properties );
1252         }
1253         long random_number_seed_for_fitch_parsimony = 0l;
1254         boolean radomize_fitch_parsimony = false;
1255         if ( cla.isOptionSet( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
1256             if ( !cla.isOptionValueSet( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
1257                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for random number seed: -"
1258                         + surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION + "=<seed>" );
1259             }
1260             try {
1261                 random_number_seed_for_fitch_parsimony = cla
1262                         .getOptionValueAsLong( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
1263             }
1264             catch ( final IOException e ) {
1265                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1266             }
1267             radomize_fitch_parsimony = true;
1268         }
1269         SortedSet<DomainId> filter = null;
1270         if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
1271                 || ( negative_domains_filter_file != null ) ) {
1272             filter = new TreeSet<DomainId>();
1273             if ( positive_filter_file != null ) {
1274                 processFilter( positive_filter_file, filter );
1275             }
1276             else if ( negative_filter_file != null ) {
1277                 processFilter( negative_filter_file, filter );
1278             }
1279             else if ( negative_domains_filter_file != null ) {
1280                 processFilter( negative_domains_filter_file, filter );
1281             }
1282         }
1283         Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps = null;
1284         File[] secondary_features_map_files = null;
1285         final File domain_lengths_analysis_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1286                 + DOMAIN_LENGTHS_ANALYSIS_SUFFIX );
1287         if ( PERFORM_DOMAIN_LENGTH_ANALYSIS ) {
1288             SurfacingUtil.checkForOutputFileWriteability( domain_lengths_analysis_outfile );
1289         }
1290         if ( cla.isOptionSet( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
1291             if ( !cla.isOptionValueSet( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
1292                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for secondary features map file: -"
1293                         + surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE + "=<file>" );
1294             }
1295             final String[] secondary_features_map_files_strs = cla
1296                     .getOptionValue( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ).split( "#" );
1297             secondary_features_map_files = new File[ secondary_features_map_files_strs.length ];
1298             domain_id_to_secondary_features_maps = new Map[ secondary_features_map_files_strs.length ];
1299             int i = 0;
1300             for( final String secondary_features_map_files_str : secondary_features_map_files_strs ) {
1301                 secondary_features_map_files[ i ] = new File( secondary_features_map_files_str );
1302                 final String error = ForesterUtil.isReadableFile( secondary_features_map_files[ i ] );
1303                 if ( !ForesterUtil.isEmpty( error ) ) {
1304                     ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read secondary features map file: " + error );
1305                 }
1306                 try {
1307                     domain_id_to_secondary_features_maps[ i ] = SurfacingUtil
1308                             .createDomainIdToSecondaryFeaturesMap( secondary_features_map_files[ i ] );
1309                 }
1310                 catch ( final IOException e ) {
1311                     ForesterUtil.fatalError( surfacing.PRG_NAME,
1312                                              "cannot read secondary features map file: " + e.getMessage() );
1313                 }
1314                 catch ( final Exception e ) {
1315                     ForesterUtil.fatalError( surfacing.PRG_NAME, "problem with contents of features map file ["
1316                             + secondary_features_map_files[ i ] + "]: " + e.getMessage() );
1317                 }
1318                 i++;
1319             }
1320         }
1321         if ( out_dir == null ) {
1322             ForesterUtil.fatalError( surfacing.PRG_NAME, "no output directory indicated (-"
1323                     + surfacing.OUTPUT_DIR_OPTION + "=<dir>)" );
1324         }
1325         if ( output_file == null ) {
1326             ForesterUtil.fatalError( surfacing.PRG_NAME, "no name for (main) output file indicated (-"
1327                     + surfacing.OUTPUT_FILE_OPTION + "=<file>)" );
1328         }
1329         if ( ( domain_id_to_go_ids_map == null ) || domain_id_to_go_ids_map.isEmpty() ) {
1330             ForesterUtil.fatalError( surfacing.PRG_NAME,
1331                                      "no (acceptable) Pfam to GO id mapping file provided ('pfam2go file') (-"
1332                                              + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>)" );
1333         }
1334         if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
1335             ForesterUtil.fatalError( surfacing.PRG_NAME,
1336                                      "no (acceptable) go id to term mapping file provided ('GO OBO file') (-"
1337                                              + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>)" );
1338         }
1339         System.out.println( "Output directory            : " + out_dir );
1340         if ( input_file_names_from_file != null ) {
1341             System.out.println( "Input files names from      : " + input_files_file + " ["
1342                     + input_file_names_from_file.length + " input files]" );
1343             html_desc.append( "<tr><td>Input files names from:</td><td>" + input_files_file + " ["
1344                     + input_file_names_from_file.length + " input files]</td></tr>" + nl );
1345         }
1346         if ( positive_filter_file != null ) {
1347             final int filter_size = filter.size();
1348             System.out.println( "Positive protein filter     : " + positive_filter_file + " [" + filter_size
1349                     + " domain ids]" );
1350             html_desc.append( "<tr><td>Positive protein filter:</td><td>" + positive_filter_file + " [" + filter_size
1351                     + " domain ids]</td></tr>" + nl );
1352         }
1353         if ( negative_filter_file != null ) {
1354             final int filter_size = filter.size();
1355             System.out.println( "Negative protein filter     : " + negative_filter_file + " [" + filter_size
1356                     + " domain ids]" );
1357             html_desc.append( "<tr><td>Negative protein filter:</td><td>" + negative_filter_file + " [" + filter_size
1358                     + " domain ids]</td></tr>" + nl );
1359         }
1360         if ( negative_domains_filter_file != null ) {
1361             final int filter_size = filter.size();
1362             System.out.println( "Negative domain filter      : " + negative_domains_filter_file + " [" + filter_size
1363                     + " domain ids]" );
1364             html_desc.append( "<tr><td>Negative domain filter:</td><td>" + negative_domains_filter_file + " ["
1365                     + filter_size + " domain ids]</td></tr>" + nl );
1366         }
1367         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
1368             String plus0 = "";
1369             for( final String s : plus_minus_analysis_high_copy_base_species ) {
1370                 plus0 += "+" + s + " ";
1371             }
1372             String plus1 = "";
1373             for( final String s : plus_minus_analysis_high_copy_target_species ) {
1374                 plus1 += "*" + s + " ";
1375             }
1376             String minus = "";
1377             for( final String s : plus_minus_analysis_high_low_copy_species ) {
1378                 minus += "-" + s + " ";
1379             }
1380             System.out.println( "Plus-minus analysis         : " + plus1 + "&& " + plus0 + "&& " + minus );
1381             html_desc.append( "<tr><td>Plus-minus analysis:</td><td>" + plus1 + "&& " + plus0 + "&& " + minus
1382                     + "</td></tr>" + nl );
1383         }
1384         if ( cutoff_scores_file != null ) {
1385             System.out.println( "Cutoff scores file          : " + cutoff_scores_file );
1386             html_desc.append( "<tr><td>Cutoff scores file:</td><td>" + cutoff_scores_file + "</td></tr>" + nl );
1387         }
1388         if ( e_value_max >= 0.0 ) {
1389             System.out.println( "E-value maximum (inclusive) : " + e_value_max );
1390             html_desc.append( "<tr><td>E-value maximum (inclusive):</td><td>" + e_value_max + "</td></tr>" + nl );
1391         }
1392         System.out.println( "Ignore DUFs                 : " + ignore_dufs );
1393         if ( ignore_virus_like_ids ) {
1394             System.out.println( "Ignore virus like ids       : " + ignore_virus_like_ids );
1395             html_desc.append( "<tr><td>Ignore virus, phage, transposition related ids:</td><td>"
1396                     + ignore_virus_like_ids + "</td></tr>" + nl );
1397         }
1398         html_desc.append( "<tr><td>Ignore DUFs:</td><td>" + ignore_dufs + "</td></tr>" + nl );
1399         if ( max_allowed_overlap != surfacing.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1400             System.out.println( "Max allowed domain overlap  : " + max_allowed_overlap );
1401             html_desc.append( "<tr><td>Max allowed domain overlap:</td><td>" + max_allowed_overlap + "</td></tr>" + nl );
1402         }
1403         if ( no_engulfing_overlaps ) {
1404             System.out.println( "Ignore engulfed domains     : " + no_engulfing_overlaps );
1405             html_desc.append( "<tr><td>Ignore (lower confidence) engulfed domains:</td><td>" + no_engulfing_overlaps
1406                     + "</td></tr>" + nl );
1407         }
1408         System.out.println( "Ignore singlet domains      : " + ignore_domains_without_combs_in_all_spec );
1409         html_desc
1410                 .append( "<tr><td>Ignore singlet domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1411                         + ignore_domains_without_combs_in_all_spec + "</td></tr>" + nl );
1412         System.out.println( "Ignore species specific doms: " + ignore_species_specific_domains );
1413         html_desc
1414                 .append( "<tr><td>Ignore species specific domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1415                         + ignore_species_specific_domains + "</td></tr>" + nl );
1416         System.out.println( "Ignore combination with self: " + ignore_combination_with_same );
1417         html_desc.append( "<tr><td>Ignore combination with self for domain combination similarity analyses:</td><td>"
1418                 + ignore_combination_with_same + "</td></tr>" + nl );
1419         System.out.println( "Consider directedness       : "
1420                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) );
1421         html_desc.append( "<tr><td>Consider directedness of binary domain combinations:</td><td>"
1422                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) + "</td></tr>" + nl );
1423         if ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) {
1424             System.out.println( "Consider adjacency          : "
1425                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) );
1426             html_desc.append( "<tr><td>Consider djacency of binary domain combinations:</td><td>"
1427                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) + "</td></tr>"
1428                     + nl );
1429         }
1430         System.out.print( "Domain counts sort order    : " );
1431         switch ( dc_sort_order ) {
1432             case ALPHABETICAL_KEY_ID:
1433                 System.out.println( "alphabetical" );
1434                 break;
1435             case KEY_DOMAIN_COUNT:
1436                 System.out.println( "domain count" );
1437                 break;
1438             case KEY_DOMAIN_PROTEINS_COUNT:
1439                 System.out.println( "domain proteins count" );
1440                 break;
1441             case COMBINATIONS_COUNT:
1442                 System.out.println( "domain combinations count" );
1443                 break;
1444             default:
1445                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for dc sort order" );
1446         }
1447         if ( domain_id_to_go_ids_map != null ) {
1448             System.out.println( "Pfam to GO mappings from    : " + pfam_to_go_file + " [" + domain_id_to_go_ids_count
1449                     + " mappings]" );
1450             html_desc.append( "<tr><td>Pfam to GO mappings from:</td><td>" + pfam_to_go_file + " ["
1451                     + domain_id_to_go_ids_count + " mappings]" + "</td></tr>" + nl );
1452         }
1453         if ( go_terms != null ) {
1454             System.out.println( "GO terms from               : " + go_obo_file + " [" + go_terms.size() + " terms]" );
1455             html_desc.append( "<tr><td>GO terms from:</td><td>" + go_obo_file + " [" + go_terms.size() + " terms]"
1456                     + "</td></tr>" + nl );
1457         }
1458         if ( go_namespace_limit != null ) {
1459             System.out.println( "Limit GO terms to           : " + go_namespace_limit.toString() );
1460             html_desc.append( "<tr><td>Limit GO terms to</td><td>" + go_namespace_limit + "</td></tr>" + nl );
1461         }
1462         if ( perform_pwc ) {
1463             System.out.println( "Suffix for PWC files        : " + automated_pairwise_comparison_suffix );
1464             html_desc.append( "<tr><td>Suffix for PWC files</td><td>" + automated_pairwise_comparison_suffix
1465                     + "</td></tr>" + nl );
1466         }
1467         if ( out_dir != null ) {
1468             System.out.println( "Output directory            : " + out_dir );
1469         }
1470         if ( query_domain_ids != null ) {
1471             System.out.println( "Query domains (ordered)     : " + query_domain_ids );
1472             html_desc.append( "<tr><td></td><td>" + query_domain_ids + "</td></tr>" + nl );
1473         }
1474         System.out.println( "Write similarities to       : " + output_file );
1475         System.out.print( "  Scoring method            : " );
1476         html_desc.append( "<tr><td>Scoring method:</td><td>" );
1477         switch ( scoring ) {
1478             case COMBINATIONS:
1479                 System.out.println( "domain combinations based" );
1480                 html_desc.append( "domain combinations based" + "</td></tr>" + nl );
1481                 break;
1482             case DOMAINS:
1483                 System.out.println( "domain counts based" );
1484                 html_desc.append( "domain counts based" + "</td></tr>" + nl );
1485                 break;
1486             case PROTEINS:
1487                 System.out.println( "domain proteins counts based" );
1488                 html_desc.append( "domain proteins counts based" + "</td></tr>" + nl );
1489                 break;
1490             default:
1491                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for scoring" );
1492         }
1493         System.out.print( "  Sort by                   : " );
1494         html_desc.append( "<tr><td>Sort by:</td><td>" );
1495         switch ( domain_similarity_sort_field ) {
1496             case MIN:
1497                 System.out.print( "score minimum" );
1498                 html_desc.append( "score minimum" );
1499                 break;
1500             case MAX:
1501                 System.out.print( "score maximum" );
1502                 html_desc.append( "score maximum" );
1503                 break;
1504             case MEAN:
1505                 System.out.print( "score mean" );
1506                 html_desc.append( "score mean" );
1507                 break;
1508             case SD:
1509                 System.out.print( "score standard deviation" );
1510                 html_desc.append( "score standard deviation" );
1511                 break;
1512             case SPECIES_COUNT:
1513                 System.out.print( "species number" );
1514                 html_desc.append( "species number" );
1515                 break;
1516             case DOMAIN_ID:
1517                 System.out.print( "alphabetical domain identifier" );
1518                 html_desc.append( "alphabetical domain identifier" );
1519                 break;
1520             case MAX_DIFFERENCE:
1521                 System.out.print( "(maximal) difference" );
1522                 html_desc.append( "(maximal) difference" );
1523                 break;
1524             case ABS_MAX_COUNTS_DIFFERENCE:
1525                 System.out.print( "absolute (maximal) counts difference" );
1526                 html_desc.append( "absolute (maximal) counts difference" );
1527                 break;
1528             case MAX_COUNTS_DIFFERENCE:
1529                 System.out.print( "(maximal) counts difference" );
1530                 html_desc.append( "(maximal) counts  difference" );
1531                 break;
1532             default:
1533                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for similarities" );
1534         }
1535         if ( sort_by_species_count_first ) {
1536             System.out.println( " (sort by species count first)" );
1537             html_desc.append( " (sort by species count first)" );
1538         }
1539         else {
1540             System.out.println();
1541         }
1542         html_desc.append( "</td></tr>" + nl );
1543         System.out.print( "  Detailedness              : " );
1544         switch ( detailedness ) {
1545             case BASIC:
1546                 System.out.println( "basic" );
1547                 break;
1548             case LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES:
1549                 System.out.println( "list combining domains for each species" );
1550                 break;
1551             case PUNCTILIOUS:
1552                 System.out.println( "punctilious" );
1553                 break;
1554             default:
1555                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for detailedness" );
1556         }
1557         System.out.print( "  Print option              : " );
1558         switch ( domain_similarity_print_option ) {
1559             case HTML:
1560                 System.out.println( "HTML" );
1561                 break;
1562             case SIMPLE_TAB_DELIMITED:
1563                 System.out.println( "simple tab delimited" );
1564                 break;
1565             default:
1566                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for print option" );
1567         }
1568         System.out.print( "  Species matrix            : " + species_matrix );
1569         System.out.println();
1570         final File dc_data_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file + DATA_FILE_SUFFIX );
1571         System.out.println( "Domain comb data output     : " + dc_data_file );
1572         html_desc.append( "<tr><td>Domain combination data output:</td><td> " + dc_data_file + " </td></tr>" );
1573         System.out.println();
1574         if ( perform_pwc ) {
1575             System.out.println( "Pairwise comparisons: " );
1576             html_desc.append( "<tr><td>Pairwise comparisons:</td><td></td></tr>" );
1577             System.out.print( "  Sort by                   : " );
1578             html_desc.append( "<tr><td>Sort by:</td><td>" );
1579             switch ( domain_similarity_sort_field_for_automated_pwc ) {
1580                 case MEAN:
1581                     System.out.print( "score mean" );
1582                     html_desc.append( "score mean" );
1583                     break;
1584                 case DOMAIN_ID:
1585                     System.out.print( "alphabetical domain identifier" );
1586                     html_desc.append( "alphabetical domain identifier" );
1587                     break;
1588                 case MAX_DIFFERENCE:
1589                     System.out.print( "difference" );
1590                     html_desc.append( "difference" );
1591                     break;
1592                 case ABS_MAX_COUNTS_DIFFERENCE:
1593                     System.out.print( "absolute counts difference" );
1594                     html_desc.append( "absolute counts difference" );
1595                     break;
1596                 case MAX_COUNTS_DIFFERENCE:
1597                     System.out.print( "counts difference" );
1598                     html_desc.append( "counts difference" );
1599                     break;
1600                 default:
1601                     ForesterUtil
1602                             .unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for similarities" );
1603             }
1604             System.out.println();
1605             html_desc.append( "</td></tr>" + nl );
1606             if ( jacknifed_distances ) {
1607                 html_desc.append( "<tr><td>Jacknife:</td><td>" + jacknife_resamplings + " resamplings</td></tr>" + nl );
1608                 html_desc.append( "<tr><td>Jacknife ratio:</td><td>" + ForesterUtil.round( jacknife_ratio, 2 )
1609                         + "</td></tr>" + nl );
1610                 html_desc.append( "<tr><td>Jacknife random number seed:</td><td>" + random_seed + "</td></tr>" + nl );
1611                 System.out.println( "  Jacknife                  : " + jacknife_resamplings + " resamplings" );
1612                 System.out.println( "    Ratio                   : " + ForesterUtil.round( jacknife_ratio, 2 ) );
1613                 System.out.println( "    Random number seed      : " + random_seed );
1614             }
1615             //                if ( infer_species_trees ) {
1616             //                    html_desc.append( "<tr><td>Infer species trees:</td><td>true</td></tr>" + nl );
1617             //                    System.out.println( "  Infer species trees       : true" );
1618             //                }
1619             if ( ( intrees != null ) && ( intrees.length > 0 ) ) {
1620                 for( final File intree_file : intree_files ) {
1621                     html_desc.append( "<tr><td>Intree for gain/loss parsimony analysis:</td><td>" + intree_file
1622                             + "</td></tr>" + nl );
1623                     System.out.println( "  Intree for gain/loss pars.: " + intree_file );
1624                 }
1625             }
1626             if ( radomize_fitch_parsimony ) {
1627                 html_desc.append( "<tr><td>    Random number seed for Fitch parsimony analysis:</td><td>"
1628                         + random_number_seed_for_fitch_parsimony + "</td></tr>" + nl );
1629                 System.out.println( "    Random number seed      : " + random_number_seed_for_fitch_parsimony );
1630             }
1631             if ( ( domain_id_to_secondary_features_maps != null ) && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
1632                 for( int i = 0; i < secondary_features_map_files.length; i++ ) {
1633                     html_desc.append( "<tr><td>Secondary features map file:</td><td>"
1634                             + secondary_features_map_files[ i ] + "</td></tr>" + nl );
1635                     System.out.println( "Secondary features map file : " + secondary_features_map_files[ i ]
1636                             + " [mappings for " + domain_id_to_secondary_features_maps[ i ].size() + " domain ids]" );
1637                     if ( VERBOSE ) {
1638                         System.out.println();
1639                         System.out.println( "Domain ids to secondary features map:" );
1640                         for( final DomainId domain_id : domain_id_to_secondary_features_maps[ i ].keySet() ) {
1641                             System.out.print( domain_id.getId() );
1642                             System.out.print( " => " );
1643                             for( final String sec : domain_id_to_secondary_features_maps[ i ].get( domain_id ) ) {
1644                                 System.out.print( sec );
1645                                 System.out.print( " " );
1646                             }
1647                             System.out.println();
1648                         }
1649                     }
1650                 }
1651             }
1652         } // if ( perform_pwc ) {
1653         System.out.println();
1654         html_desc.append( "<tr><td>Command line:</td><td>\n" + cla.getCommandLineArgsAsString() + "\n</td></tr>" + nl );
1655         System.out.println( "Command line                : " + cla.getCommandLineArgsAsString() );
1656         BufferedWriter[] query_domains_writer_ary = null;
1657         List<DomainId>[] query_domain_ids_array = null;
1658         if ( query_domain_ids != null ) {
1659             final String[] query_domain_ids_str_array = query_domain_ids.split( "#" );
1660             query_domain_ids_array = new ArrayList[ query_domain_ids_str_array.length ];
1661             query_domains_writer_ary = new BufferedWriter[ query_domain_ids_str_array.length ];
1662             for( int i = 0; i < query_domain_ids_str_array.length; i++ ) {
1663                 String query_domain_ids_str = query_domain_ids_str_array[ i ];
1664                 final String[] query_domain_ids_str_ary = query_domain_ids_str.split( "~" );
1665                 final List<DomainId> query = new ArrayList<DomainId>();
1666                 for( final String element : query_domain_ids_str_ary ) {
1667                     query.add( new DomainId( element ) );
1668                 }
1669                 query_domain_ids_array[ i ] = query;
1670                 query_domain_ids_str = query_domain_ids_str.replace( '~', '_' );
1671                 String protein_names_writer_str = query_domain_ids_str + surfacing.SEQ_EXTRACT_SUFFIX;
1672                 if ( out_dir != null ) {
1673                     protein_names_writer_str = out_dir + ForesterUtil.FILE_SEPARATOR + protein_names_writer_str;
1674                 }
1675                 try {
1676                     query_domains_writer_ary[ i ] = new BufferedWriter( new FileWriter( protein_names_writer_str ) );
1677                 }
1678                 catch ( final IOException e ) {
1679                     ForesterUtil.fatalError( surfacing.PRG_NAME, "Could not open [" + protein_names_writer_str + "]: "
1680                             + e.getLocalizedMessage() );
1681                 }
1682             }
1683         }
1684         SortedMap<Species, List<Protein>> protein_lists_per_species = null; //This will only be created if neede.
1685         boolean need_protein_lists_per_species = false;
1686         if ( ( plus_minus_analysis_high_copy_base_species.size() > 0 ) || output_protein_lists_for_all_domains ) {
1687             need_protein_lists_per_species = true;
1688         }
1689         if ( need_protein_lists_per_species ) {
1690             protein_lists_per_species = new TreeMap<Species, List<Protein>>();
1691         }
1692         final List<GenomeWideCombinableDomains> gwcd_list = new ArrayList<GenomeWideCombinableDomains>( number_of_genomes );
1693         final SortedSet<DomainId> all_domains_encountered = new TreeSet<DomainId>();
1694         final SortedSet<BinaryDomainCombination> all_bin_domain_combinations_encountered = new TreeSet<BinaryDomainCombination>();
1695         List<BinaryDomainCombination> all_bin_domain_combinations_gained_fitch = null;
1696         List<BinaryDomainCombination> all_bin_domain_combinations_lost_fitch = null;
1697         if ( ( intrees != null ) && ( intrees.length == 1 ) ) {
1698             all_bin_domain_combinations_gained_fitch = new ArrayList<BinaryDomainCombination>();
1699             all_bin_domain_combinations_lost_fitch = new ArrayList<BinaryDomainCombination>();
1700         }
1701         final DomainLengthsTable domain_lengths_table = new DomainLengthsTable();
1702         final File per_genome_domain_promiscuity_statistics_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR
1703                 + output_file + D_PROMISCUITY_FILE_SUFFIX );
1704         BufferedWriter per_genome_domain_promiscuity_statistics_writer = null;
1705         try {
1706             per_genome_domain_promiscuity_statistics_writer = new BufferedWriter( new FileWriter( per_genome_domain_promiscuity_statistics_file ) );
1707             per_genome_domain_promiscuity_statistics_writer.write( "Species:\t" );
1708             per_genome_domain_promiscuity_statistics_writer.write( "Mean:\t" );
1709             per_genome_domain_promiscuity_statistics_writer.write( "SD:\t" );
1710             per_genome_domain_promiscuity_statistics_writer.write( "Median:\t" );
1711             per_genome_domain_promiscuity_statistics_writer.write( "Min:\t" );
1712             per_genome_domain_promiscuity_statistics_writer.write( "Max:\t" );
1713             per_genome_domain_promiscuity_statistics_writer.write( "N:\t" );
1714             per_genome_domain_promiscuity_statistics_writer.write( "Max Promiscuous Domains:"
1715                     + ForesterUtil.LINE_SEPARATOR );
1716         }
1717         catch ( final IOException e2 ) {
1718             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1719         }
1720         final File log_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file + LOG_FILE_SUFFIX );
1721         BufferedWriter log_writer = null;
1722         try {
1723             log_writer = new BufferedWriter( new FileWriter( log_file ) );
1724         }
1725         catch ( final IOException e2 ) {
1726             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1727         }
1728         BufferedWriter dc_data_writer = null;
1729         try {
1730             dc_data_writer = new BufferedWriter( new FileWriter( dc_data_file ) );
1731             dc_data_writer.write( DATA_FILE_DESC );
1732             dc_data_writer.write( ForesterUtil.LINE_SEPARATOR );
1733         }
1734         catch ( final IOException e2 ) {
1735             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1736         }
1737         final DescriptiveStatistics protein_coverage_stats = new BasicDescriptiveStatistics();
1738         final DescriptiveStatistics all_genomes_domains_per_potein_stats = new BasicDescriptiveStatistics();
1739         final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo = new TreeMap<Integer, Integer>();
1740         final SortedSet<String> domains_which_are_always_single = new TreeSet<String>();
1741         final SortedSet<String> domains_which_are_sometimes_single_sometimes_not = new TreeSet<String>();
1742         final SortedSet<String> domains_which_never_single = new TreeSet<String>();
1743         BufferedWriter domains_per_potein_stats_writer = null;
1744         try {
1745             domains_per_potein_stats_writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR
1746                     + output_file + "__domains_per_potein_stats.txt" ) );
1747             domains_per_potein_stats_writer.write( "Genome" );
1748             domains_per_potein_stats_writer.write( "\t" );
1749             domains_per_potein_stats_writer.write( "Mean" );
1750             domains_per_potein_stats_writer.write( "\t" );
1751             domains_per_potein_stats_writer.write( "SD" );
1752             domains_per_potein_stats_writer.write( "\t" );
1753             domains_per_potein_stats_writer.write( "Median" );
1754             domains_per_potein_stats_writer.write( "\t" );
1755             domains_per_potein_stats_writer.write( "N" );
1756             domains_per_potein_stats_writer.write( "\t" );
1757             domains_per_potein_stats_writer.write( "Min" );
1758             domains_per_potein_stats_writer.write( "\t" );
1759             domains_per_potein_stats_writer.write( "Max" );
1760             domains_per_potein_stats_writer.write( "\n" );
1761         }
1762         catch ( final IOException e3 ) {
1763             e3.printStackTrace();
1764         }
1765         Map<String, DescriptiveStatistics> protein_length_stats_by_dc = null;
1766         Map<String, DescriptiveStatistics> domain_number_stats_by_dc = null;
1767         final Map<String, DescriptiveStatistics> domain_length_stats_by_domain = new HashMap<String, DescriptiveStatistics>();
1768         if ( PERFORM_DC_REGAIN_PROTEINS_STATS ) {
1769             protein_length_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
1770             domain_number_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
1771         }
1772         // Main loop:
1773         for( int i = 0; i < number_of_genomes; ++i ) {
1774             System.out.println();
1775             System.out.println( ( i + 1 ) + "/" + number_of_genomes );
1776             log( ( i + 1 ) + "/" + number_of_genomes, log_writer );
1777             System.out.println( "Processing                                     : " + input_file_properties[ i ][ 0 ] );
1778             log( "Genome                                         : " + input_file_properties[ i ][ 0 ], log_writer );
1779             HmmscanPerDomainTableParser parser = null;
1780             INDIVIDUAL_SCORE_CUTOFF ind_score_cutoff = INDIVIDUAL_SCORE_CUTOFF.NONE;
1781             if ( individual_score_cutoffs != null ) {
1782                 ind_score_cutoff = INDIVIDUAL_SCORE_CUTOFF_DEFAULT;
1783             }
1784             if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
1785                     || ( negative_domains_filter_file != null ) ) {
1786                 HmmscanPerDomainTableParser.FilterType filter_type = HmmscanPerDomainTableParser.FilterType.NONE;
1787                 if ( positive_filter_file != null ) {
1788                     filter_type = HmmscanPerDomainTableParser.FilterType.POSITIVE_PROTEIN;
1789                 }
1790                 else if ( negative_filter_file != null ) {
1791                     filter_type = HmmscanPerDomainTableParser.FilterType.NEGATIVE_PROTEIN;
1792                 }
1793                 else if ( negative_domains_filter_file != null ) {
1794                     filter_type = HmmscanPerDomainTableParser.FilterType.NEGATIVE_DOMAIN;
1795                 }
1796                 parser = new HmmscanPerDomainTableParser( new File( input_file_properties[ i ][ 0 ] ),
1797                                                           input_file_properties[ i ][ 1 ],
1798                                                           filter,
1799                                                           filter_type,
1800                                                           ind_score_cutoff,
1801                                                           true );
1802             }
1803             else {
1804                 parser = new HmmscanPerDomainTableParser( new File( input_file_properties[ i ][ 0 ] ),
1805                                                           input_file_properties[ i ][ 1 ],
1806                                                           ind_score_cutoff,
1807                                                           true );
1808             }
1809             if ( e_value_max >= 0.0 ) {
1810                 parser.setEValueMaximum( e_value_max );
1811             }
1812             parser.setIgnoreDufs( ignore_dufs );
1813             parser.setIgnoreVirusLikeIds( ignore_virus_like_ids );
1814             parser.setIgnoreEngulfedDomains( no_engulfing_overlaps );
1815             if ( max_allowed_overlap != surfacing.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1816                 parser.setMaxAllowedOverlap( max_allowed_overlap );
1817             }
1818             parser.setReturnType( HmmscanPerDomainTableParser.ReturnType.UNORDERED_PROTEIN_DOMAIN_COLLECTION_PER_PROTEIN );
1819             if ( individual_score_cutoffs != null ) {
1820                 parser.setIndividualScoreCutoffs( individual_score_cutoffs );
1821             }
1822             List<Protein> protein_list = null;
1823             try {
1824                 protein_list = parser.parse();
1825             }
1826             catch ( final IOException e ) {
1827                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1828             }
1829             catch ( final Exception e ) {
1830                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, e.getMessage(), e );
1831             }
1832             if ( VERBOSE ) {
1833                 System.out.println( "Domains ignored due to negative domain filter: " );
1834                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToNegativeDomainFilterCountsMap() );
1835                 System.out.println( "Domains ignored due to virus like id: " );
1836                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToVirusLikeIdCountsMap() );
1837             }
1838             final double coverage = ( double ) protein_list.size() / parser.getProteinsEncountered();
1839             protein_coverage_stats.addValue( coverage );
1840             System.out.println( "Number of proteins encountered                 : " + parser.getProteinsEncountered() );
1841             log( "Number of proteins encountered                 : " + parser.getProteinsEncountered(), log_writer );
1842             System.out.println( "Number of proteins stored                      : " + protein_list.size() );
1843             log( "Number of proteins stored                      : " + protein_list.size(), log_writer );
1844             System.out.println( "Coverage                                       : "
1845                     + ForesterUtil.roundToInt( 100.0 * coverage ) + "%" );
1846             log( "Coverage                                       : " + ForesterUtil.roundToInt( 100.0 * coverage )
1847                     + "%", log_writer );
1848             System.out.println( "Domains encountered                            : " + parser.getDomainsEncountered() );
1849             log( "Domains encountered                            : " + parser.getDomainsEncountered(), log_writer );
1850             System.out.println( "Domains stored                                 : " + parser.getDomainsStored() );
1851             log( "Domains stored                                 : " + parser.getDomainsStored(), log_writer );
1852             System.out.println( "Distinct domains stored                        : "
1853                     + parser.getDomainsStoredSet().size() );
1854             log( "Distinct domains stored                        : " + parser.getDomainsStoredSet().size(), log_writer );
1855             System.out.println( "Domains ignored due to individual score cutoffs: "
1856                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff() );
1857             log( "Domains ignored due to individual score cutoffs: "
1858                          + parser.getDomainsIgnoredDueToIndividualScoreCutoff(),
1859                  log_writer );
1860             System.out.println( "Domains ignored due to E-value                 : "
1861                     + parser.getDomainsIgnoredDueToEval() );
1862             log( "Domains ignored due to E-value                 : " + parser.getDomainsIgnoredDueToEval(), log_writer );
1863             System.out.println( "Domains ignored due to DUF designation         : "
1864                     + parser.getDomainsIgnoredDueToDuf() );
1865             log( "Domains ignored due to DUF designation         : " + parser.getDomainsIgnoredDueToDuf(), log_writer );
1866             if ( ignore_virus_like_ids ) {
1867                 System.out.println( "Domains ignored due virus like ids             : "
1868                         + parser.getDomainsIgnoredDueToVirusLikeIds() );
1869                 log( "Domains ignored due virus like ids             : " + parser.getDomainsIgnoredDueToVirusLikeIds(),
1870                      log_writer );
1871             }
1872             System.out.println( "Domains ignored due negative domain filter     : "
1873                     + parser.getDomainsIgnoredDueToNegativeDomainFilter() );
1874             log( "Domains ignored due negative domain filter     : "
1875                          + parser.getDomainsIgnoredDueToNegativeDomainFilter(),
1876                  log_writer );
1877             System.out.println( "Domains ignored due to overlap                 : "
1878                     + parser.getDomainsIgnoredDueToOverlap() );
1879             log( "Domains ignored due to overlap                 : " + parser.getDomainsIgnoredDueToOverlap(),
1880                  log_writer );
1881             if ( negative_filter_file != null ) {
1882                 System.out.println( "Proteins ignored due to negative filter        : "
1883                         + parser.getProteinsIgnoredDueToFilter() );
1884                 log( "Proteins ignored due to negative filter        : " + parser.getProteinsIgnoredDueToFilter(),
1885                      log_writer );
1886             }
1887             if ( positive_filter_file != null ) {
1888                 System.out.println( "Proteins ignored due to positive filter        : "
1889                         + parser.getProteinsIgnoredDueToFilter() );
1890                 log( "Proteins ignored due to positive filter        : " + parser.getProteinsIgnoredDueToFilter(),
1891                      log_writer );
1892             }
1893             System.out.println( "Time for processing                            : " + parser.getTime() + "ms" );
1894             log( "", log_writer );
1895             html_desc.append( "<tr><td>" + input_file_properties[ i ][ 0 ] + " [species: "
1896                     + input_file_properties[ i ][ 1 ] + "]" + ":</td><td>domains analyzed: "
1897                     + parser.getDomainsStored() + "; domains ignored: [ind score cutoffs: "
1898                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff() + "] [E-value cutoff: "
1899                     + parser.getDomainsIgnoredDueToEval() + "] [DUF: " + parser.getDomainsIgnoredDueToDuf()
1900                     + "] [virus like ids: " + parser.getDomainsIgnoredDueToVirusLikeIds()
1901                     + "] [negative domain filter: " + parser.getDomainsIgnoredDueToNegativeDomainFilter()
1902                     + "] [overlap: " + parser.getDomainsIgnoredDueToOverlap() + "]" );
1903             if ( negative_filter_file != null ) {
1904                 html_desc.append( "; proteins ignored due to negative filter: "
1905                         + parser.getProteinsIgnoredDueToFilter() );
1906             }
1907             if ( positive_filter_file != null ) {
1908                 html_desc.append( "; proteins ignored due to positive filter: "
1909                         + parser.getProteinsIgnoredDueToFilter() );
1910             }
1911             html_desc.append( "</td></tr>" + nl );
1912             // domain_partner_counts_array[ i ] =
1913             // Methods.getDomainPartnerCounts( protein_domain_collections_array[
1914             // i ],
1915             // false, input_file_properties[ i ][ 1 ] );
1916             try {
1917                 int count = 0;
1918                 for( final Protein protein : protein_list ) {
1919                     dc_data_writer.write( SurfacingUtil.proteinToDomainCombinations( protein, count + "", "\t" )
1920                             .toString() );
1921                     ++count;
1922                     for( final Domain d : protein.getProteinDomains() ) {
1923                         final String d_str = d.getDomainId().toString();
1924                         if ( !domain_length_stats_by_domain.containsKey( d_str ) ) {
1925                             domain_length_stats_by_domain.put( d_str, new BasicDescriptiveStatistics() );
1926                         }
1927                         domain_length_stats_by_domain.get( d_str ).addValue( d.getLength() );
1928                     }
1929                 }
1930             }
1931             catch ( final IOException e ) {
1932                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.toString() );
1933             }
1934             SurfacingUtil.domainsPerProteinsStatistics( input_file_properties[ i ][ 1 ],
1935                                                         protein_list,
1936                                                         all_genomes_domains_per_potein_stats,
1937                                                         all_genomes_domains_per_potein_histo,
1938                                                         domains_which_are_always_single,
1939                                                         domains_which_are_sometimes_single_sometimes_not,
1940                                                         domains_which_never_single,
1941                                                         domains_per_potein_stats_writer );
1942             gwcd_list.add( BasicGenomeWideCombinableDomains
1943                     .createInstance( protein_list,
1944                                      ignore_combination_with_same,
1945                                      new BasicSpecies( input_file_properties[ i ][ 1 ] ),
1946                                      domain_id_to_go_ids_map,
1947                                      dc_type,
1948                                      protein_length_stats_by_dc,
1949                                      domain_number_stats_by_dc ) );
1950             domain_lengths_table.addLengths( protein_list );
1951             if ( gwcd_list.get( i ).getSize() > 0 ) {
1952                 SurfacingUtil.writeDomainCombinationsCountsFile( input_file_properties,
1953                                                                  out_dir,
1954                                                                  per_genome_domain_promiscuity_statistics_writer,
1955                                                                  gwcd_list.get( i ),
1956                                                                  i,
1957                                                                  dc_sort_order );
1958                 if ( output_binary_domain_combinationsfor_graph_analysis ) {
1959                     SurfacingUtil.writeBinaryDomainCombinationsFileForGraphAnalysis( input_file_properties,
1960                                                                                      out_dir,
1961                                                                                      gwcd_list.get( i ),
1962                                                                                      i,
1963                                                                                      dc_sort_order );
1964                 }
1965                 SurfacingUtil.addAllDomainIdsToSet( gwcd_list.get( i ), all_domains_encountered );
1966                 SurfacingUtil.addAllBinaryDomainCombinationToSet( gwcd_list.get( i ),
1967                                                                   all_bin_domain_combinations_encountered );
1968             }
1969             if ( query_domains_writer_ary != null ) {
1970                 for( int j = 0; j < query_domain_ids_array.length; j++ ) {
1971                     try {
1972                         SurfacingUtil.extractProteinNames( protein_list,
1973                                                            query_domain_ids_array[ j ],
1974                                                            query_domains_writer_ary[ j ],
1975                                                            "\t",
1976                                                            LIMIT_SPEC_FOR_PROT_EX );
1977                         query_domains_writer_ary[ j ].flush();
1978                     }
1979                     catch ( final IOException e ) {
1980                         e.printStackTrace();
1981                     }
1982                 }
1983             }
1984             if ( need_protein_lists_per_species ) {
1985                 protein_lists_per_species.put( new BasicSpecies( input_file_properties[ i ][ 1 ] ), protein_list );
1986             }
1987             try {
1988                 log_writer.flush();
1989             }
1990             catch ( final IOException e2 ) {
1991                 ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
1992             }
1993             System.gc();
1994         } // for( int i = 0; i < number_of_genomes; ++i ) {
1995         ForesterUtil.programMessage( PRG_NAME, "Wrote domain promiscuities to: "
1996                 + per_genome_domain_promiscuity_statistics_file );
1997         //
1998         try {
1999             domains_per_potein_stats_writer.write( "ALL" );
2000             domains_per_potein_stats_writer.write( "\t" );
2001             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.arithmeticMean() + "" );
2002             domains_per_potein_stats_writer.write( "\t" );
2003             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.sampleStandardDeviation() + "" );
2004             domains_per_potein_stats_writer.write( "\t" );
2005             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.median() + "" );
2006             domains_per_potein_stats_writer.write( "\t" );
2007             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getN() + "" );
2008             domains_per_potein_stats_writer.write( "\t" );
2009             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMin() + "" );
2010             domains_per_potein_stats_writer.write( "\t" );
2011             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMax() + "" );
2012             domains_per_potein_stats_writer.write( "\n" );
2013             domains_per_potein_stats_writer.close();
2014             printOutPercentageOfMultidomainProteins( all_genomes_domains_per_potein_histo, log_writer );
2015             ForesterUtil.map2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
2016                     + "__all_genomes_domains_per_potein_histo.txt" ), all_genomes_domains_per_potein_histo, "\t", "\n" );
2017             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
2018                     + "__domains_always_single_.txt" ), domains_which_are_always_single, "\n" );
2019             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
2020                     + "__domains_single_or_combined.txt" ), domains_which_are_sometimes_single_sometimes_not, "\n" );
2021             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
2022                     + "__domains_always_combined.txt" ), domains_which_never_single, "\n" );
2023             ForesterUtil.programMessage( PRG_NAME,
2024                                          "Average of proteins with a least one domain assigned: "
2025                                                  + ( 100 * protein_coverage_stats.arithmeticMean() ) + "% (+/-"
2026                                                  + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)" );
2027             ForesterUtil.programMessage( PRG_NAME, "Range of proteins with a least one domain assigned: " + 100
2028                     * protein_coverage_stats.getMin() + "%-" + 100 * protein_coverage_stats.getMax() + "%" );
2029             log( "Average of prot with a least one dom assigned  : " + ( 100 * protein_coverage_stats.arithmeticMean() )
2030                     + "% (+/-" + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)", log_writer );
2031             log( "Range of prot with a least one dom assigned    : " + 100 * protein_coverage_stats.getMin() + "%-"
2032                     + 100 * protein_coverage_stats.getMax() + "%", log_writer );
2033         }
2034         catch ( final IOException e2 ) {
2035             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
2036         }
2037         if ( query_domains_writer_ary != null ) {
2038             for( int j = 0; j < query_domain_ids_array.length; j++ ) {
2039                 try {
2040                     query_domains_writer_ary[ j ].close();
2041                 }
2042                 catch ( final IOException e ) {
2043                     ForesterUtil.fatalError( surfacing.PRG_NAME, e.toString() );
2044                 }
2045             }
2046         }
2047         try {
2048             per_genome_domain_promiscuity_statistics_writer.close();
2049             dc_data_writer.close();
2050             log_writer.close();
2051         }
2052         catch ( final IOException e2 ) {
2053             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
2054         }
2055         if ( PERFORM_DOMAIN_LENGTH_ANALYSIS ) {
2056             try {
2057                 SurfacingUtil.executeDomainLengthAnalysis( input_file_properties,
2058                                                            number_of_genomes,
2059                                                            domain_lengths_table,
2060                                                            domain_lengths_analysis_outfile );
2061             }
2062             catch ( final IOException e1 ) {
2063                 ForesterUtil.fatalError( surfacing.PRG_NAME, e1.toString() );
2064             }
2065             System.out.println();
2066             ForesterUtil.programMessage( PRG_NAME, "Wrote domain length data to: " + domain_lengths_analysis_outfile );
2067             System.out.println();
2068         }
2069         final long analysis_start_time = new Date().getTime();
2070         PairwiseDomainSimilarityCalculator pw_calc = null;
2071         // double[] values_for_all_scores_histogram = null;
2072         final DomainSimilarityCalculator calc = new BasicDomainSimilarityCalculator( domain_similarity_sort_field,
2073                                                                                      sort_by_species_count_first,
2074                                                                                      number_of_genomes == 2 );
2075         switch ( scoring ) {
2076             case COMBINATIONS:
2077                 pw_calc = new CombinationsBasedPairwiseDomainSimilarityCalculator();
2078                 break;
2079             case DOMAINS:
2080                 pw_calc = new DomainCountsBasedPairwiseSimilarityCalculator();
2081                 break;
2082             case PROTEINS:
2083                 pw_calc = new ProteinCountsBasedPairwiseDomainSimilarityCalculator();
2084                 break;
2085             default:
2086                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for scoring" );
2087         }
2088         DomainSimilarityCalculator.GoAnnotationOutput go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.NONE;
2089         if ( domain_id_to_go_ids_map != null ) {
2090             go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.ALL;
2091         }
2092         final SortedSet<DomainSimilarity> similarities = calc
2093                 .calculateSimilarities( pw_calc,
2094                                         gwcd_list,
2095                                         ignore_domains_without_combs_in_all_spec,
2096                                         ignore_species_specific_domains );
2097         SurfacingUtil.decoratePrintableDomainSimilarities( similarities,
2098                                                            detailedness,
2099                                                            go_annotation_output,
2100                                                            go_id_to_term_map,
2101                                                            go_namespace_limit );
2102         DescriptiveStatistics pw_stats = null;
2103         try {
2104             String my_outfile = output_file.toString();
2105             Map<Character, Writer> split_writers = null;
2106             Writer writer = null;
2107             if ( similarities.size() > MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING ) {
2108                 if ( my_outfile.endsWith( ".html" ) ) {
2109                     my_outfile = my_outfile.substring( 0, my_outfile.length() - 5 );
2110                 }
2111                 split_writers = new HashMap<Character, Writer>();
2112                 createSplitWriters( out_dir, my_outfile, split_writers );
2113             }
2114             else if ( !my_outfile.endsWith( ".html" ) ) {
2115                 my_outfile += ".html";
2116                 writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
2117             }
2118             List<Species> species_order = null;
2119             if ( species_matrix ) {
2120                 species_order = new ArrayList<Species>();
2121                 for( int i = 0; i < number_of_genomes; i++ ) {
2122                     species_order.add( new BasicSpecies( input_file_properties[ i ][ 1 ] ) );
2123                 }
2124             }
2125             html_desc.append( "<tr><td>Sum of all distinct binary combinations:</td><td>"
2126                     + all_bin_domain_combinations_encountered.size() + "</td></tr>" + nl );
2127             html_desc.append( "<tr><td>Sum of all distinct domains:</td><td>" + all_domains_encountered.size()
2128                     + "</td></tr>" + nl );
2129             html_desc.append( "<tr><td>Analysis date/time:</td><td>"
2130                     + new java.text.SimpleDateFormat( "yyyy.MM.dd HH:mm:ss" ).format( new java.util.Date() )
2131                     + "</td></tr>" + nl );
2132             html_desc.append( "</table>" + nl );
2133             pw_stats = SurfacingUtil
2134                     .writeDomainSimilaritiesToFile( html_desc,
2135                                                     new StringBuilder( number_of_genomes + " genomes" ),
2136                                                     writer,
2137                                                     split_writers,
2138                                                     similarities,
2139                                                     number_of_genomes == 2,
2140                                                     species_order,
2141                                                     domain_similarity_print_option,
2142                                                     domain_similarity_sort_field,
2143                                                     scoring,
2144                                                     true );
2145             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote main output (includes domain similarities) to: \""
2146                     + ( out_dir == null ? my_outfile : out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) + "\"" );
2147         }
2148         catch ( final IOException e ) {
2149             ForesterUtil.fatalError( surfacing.PRG_NAME, "Failed to write similarites to: \"" + output_file + "\" ["
2150                     + e.getMessage() + "]" );
2151         }
2152         System.out.println();
2153         // values_for_all_scores_histogram = pw_stats.getDataAsDoubleArray();
2154         final Species[] species = new Species[ number_of_genomes ];
2155         for( int i = 0; i < number_of_genomes; ++i ) {
2156             species[ i ] = new BasicSpecies( input_file_properties[ i ][ 1 ] );
2157         }
2158         List<Phylogeny> inferred_trees = null;
2159         if ( ( number_of_genomes > 2 ) && perform_pwc ) {
2160             final PairwiseGenomeComparator pwgc = new PairwiseGenomeComparator();
2161             pwgc.performPairwiseComparisons( html_desc,
2162                                              sort_by_species_count_first,
2163                                              detailedness,
2164                                              ignore_domains_without_combs_in_all_spec,
2165                                              ignore_species_specific_domains,
2166                                              domain_similarity_sort_field_for_automated_pwc,
2167                                              domain_similarity_print_option,
2168                                              scoring,
2169                                              domain_id_to_go_ids_map,
2170                                              go_id_to_term_map,
2171                                              go_namespace_limit,
2172                                              species,
2173                                              number_of_genomes,
2174                                              gwcd_list,
2175                                              pw_calc,
2176                                              automated_pairwise_comparison_suffix,
2177                                              true,
2178                                              surfacing.PAIRWISE_DOMAIN_COMPARISONS_PREFIX,
2179                                              surfacing.PRG_NAME,
2180                                              out_dir,
2181                                              write_pwc_files );
2182             String matrix_output_file = new String( output_file.toString() );
2183             if ( matrix_output_file.indexOf( '.' ) > 1 ) {
2184                 matrix_output_file = matrix_output_file.substring( 0, matrix_output_file.indexOf( '.' ) );
2185             }
2186             if ( out_dir != null ) {
2187                 matrix_output_file = out_dir + ForesterUtil.FILE_SEPARATOR + matrix_output_file;
2188                 output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
2189             }
2190             SurfacingUtil.writeMatrixToFile( new File( matrix_output_file
2191                     + surfacing.MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ), pwgc.getDomainDistanceScoresMeans() );
2192             SurfacingUtil
2193                     .writeMatrixToFile( new File( matrix_output_file
2194                                                 + surfacing.MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ),
2195                                         pwgc.getSharedBinaryCombinationsBasedDistances() );
2196             SurfacingUtil.writeMatrixToFile( new File( matrix_output_file
2197                                                      + surfacing.MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ),
2198                                              pwgc.getSharedDomainsBasedDistances() );
2199             final Phylogeny nj_gd = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2200                     + surfacing.NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ), pwgc.getDomainDistanceScoresMeans()
2201                     .get( 0 ) );
2202             final Phylogeny nj_bc = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2203                     + surfacing.NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
2204                     .getSharedBinaryCombinationsBasedDistances().get( 0 ) );
2205             final Phylogeny nj_d = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2206                     + surfacing.NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
2207                     .getSharedDomainsBasedDistances().get( 0 ) );
2208             inferred_trees = new ArrayList<Phylogeny>();
2209             inferred_trees.add( nj_gd );
2210             inferred_trees.add( nj_bc );
2211             inferred_trees.add( nj_d );
2212             if ( jacknifed_distances ) {
2213                 pwgc.performPairwiseComparisonsJacknifed( species,
2214                                                           number_of_genomes,
2215                                                           gwcd_list,
2216                                                           true,
2217                                                           jacknife_resamplings,
2218                                                           jacknife_ratio,
2219                                                           random_seed );
2220                 SurfacingUtil
2221                         .writeMatrixToFile( new File( matrix_output_file
2222                                                     + "_"
2223                                                     + ForesterUtil.round( jacknife_ratio, 2 )
2224                                                     + "_"
2225                                                     + jacknife_resamplings
2226                                                     + surfacing.MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ),
2227                                             pwgc.getSharedBinaryCombinationsBasedDistances() );
2228                 SurfacingUtil
2229                         .writeMatrixToFile( new File( matrix_output_file + "_" + ForesterUtil.round( jacknife_ratio, 2 )
2230                                                     + "_" + jacknife_resamplings
2231                                                     + surfacing.MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ),
2232                                             pwgc.getSharedDomainsBasedDistances() );
2233                 //                if ( infer_species_trees ) {
2234                 //                    inferSpeciesTrees( new File( output_file + "_" + jacknife_resamplings
2235                 //                            + INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc
2236                 //                            .getSharedBinaryCombinationsBasedDistances() );
2237                 //                    inferSpeciesTrees( new File( output_file + "_" + jacknife_resamplings
2238                 //                            + INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc.getSharedDomainsBasedDistances() );
2239                 //                }
2240             }
2241         } // if ( ( output_file != null ) && ( number_of_genomes > 2 ) && !isEmpty( automated_pairwise_comparison_suffix ) )
2242         if ( ( out_dir != null ) && ( !perform_pwc ) ) {
2243             output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
2244         }
2245         writePresentToNexus( output_file, positive_filter_file, filter, gwcd_list );
2246         if ( ( ( intrees != null ) && ( intrees.length > 0 ) ) && ( number_of_genomes > 2 ) ) {
2247             final StringBuilder parameters_sb = createParametersAsString( ignore_dufs,
2248                                                                           e_value_max,
2249                                                                           max_allowed_overlap,
2250                                                                           no_engulfing_overlaps,
2251                                                                           cutoff_scores_file,
2252                                                                           dc_type );
2253             String s = "_";
2254             if ( radomize_fitch_parsimony ) {
2255                 s += random_number_seed_for_fitch_parsimony + "_";
2256             }
2257             int i = 0;
2258             for( final Phylogeny intree : intrees ) {
2259                 final String outfile_name = ForesterUtil.removeSuffix( output_file.toString() ) + s
2260                         + ForesterUtil.removeSuffix( intree_files[ i ].toString() );
2261                 final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator.createInstance( intree,
2262                                                                                                              gwcd_list );
2263                 SurfacingUtil.executeParsimonyAnalysis( random_number_seed_for_fitch_parsimony,
2264                                                         radomize_fitch_parsimony,
2265                                                         outfile_name,
2266                                                         domain_parsimony,
2267                                                         intree,
2268                                                         domain_id_to_go_ids_map,
2269                                                         go_id_to_term_map,
2270                                                         go_namespace_limit,
2271                                                         parameters_sb.toString(),
2272                                                         domain_id_to_secondary_features_maps,
2273                                                         positive_filter_file == null ? null : filter,
2274                                                         output_binary_domain_combinationsfor_graph_analysis,
2275                                                         all_bin_domain_combinations_gained_fitch,
2276                                                         all_bin_domain_combinations_lost_fitch,
2277                                                         dc_type,
2278                                                         protein_length_stats_by_dc,
2279                                                         domain_number_stats_by_dc,
2280                                                         domain_length_stats_by_domain );
2281                 // Listing of all domain combinations gained is only done if only one input tree is used.
2282                 if ( ( domain_id_to_secondary_features_maps != null )
2283                         && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
2284                     int j = 0;
2285                     for( final Map<DomainId, Set<String>> domain_id_to_secondary_features_map : domain_id_to_secondary_features_maps ) {
2286                         final Map<Species, MappingResults> mapping_results_map = new TreeMap<Species, MappingResults>();
2287                         final DomainParsimonyCalculator secondary_features_parsimony = DomainParsimonyCalculator
2288                                 .createInstance( intree, gwcd_list, domain_id_to_secondary_features_map );
2289                         SurfacingUtil
2290                                 .executeParsimonyAnalysisForSecondaryFeatures( outfile_name
2291                                                                                        + "_"
2292                                                                                        + secondary_features_map_files[ j++ ],
2293                                                                                secondary_features_parsimony,
2294                                                                                intree,
2295                                                                                parameters_sb.toString(),
2296                                                                                mapping_results_map );
2297                         if ( i == 0 ) {
2298                             System.out.println();
2299                             System.out.println( "Mapping to secondary features:" );
2300                             for( final Species spec : mapping_results_map.keySet() ) {
2301                                 final MappingResults mapping_results = mapping_results_map.get( spec );
2302                                 final int total_domains = mapping_results.getSumOfFailures()
2303                                         + mapping_results.getSumOfSuccesses();
2304                                 System.out.print( spec + ":" );
2305                                 System.out.print( " mapped domains = " + mapping_results.getSumOfSuccesses() );
2306                                 System.out.print( ", not mapped domains = " + mapping_results.getSumOfFailures() );
2307                                 if ( total_domains > 0 ) {
2308                                     System.out.println( ", mapped ratio = "
2309                                             + ( 100 * mapping_results.getSumOfSuccesses() / total_domains ) + "%" );
2310                                 }
2311                                 else {
2312                                     System.out.println( ", mapped ratio = n/a (total domains = 0 )" );
2313                                 }
2314                             }
2315                         }
2316                     }
2317                 }
2318                 i++;
2319             } // for( final Phylogeny intree : intrees ) {
2320         }
2321         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
2322             executePlusMinusAnalysis( output_file,
2323                                       plus_minus_analysis_high_copy_base_species,
2324                                       plus_minus_analysis_high_copy_target_species,
2325                                       plus_minus_analysis_high_low_copy_species,
2326                                       gwcd_list,
2327                                       protein_lists_per_species,
2328                                       domain_id_to_go_ids_map,
2329                                       go_id_to_term_map,
2330                                       plus_minus_analysis_numbers );
2331         }
2332         if ( output_protein_lists_for_all_domains ) {
2333             writeProteinListsForAllSpecies( out_dir, protein_lists_per_species, gwcd_list );
2334         }
2335         //        if ( ( intrees != null ) && ( intrees.length > 0 ) && ( inferred_trees != null ) && ( inferred_trees.size() > 0 ) ) {
2336         //            final StringBuilder parameters_sb = createParametersAsString( ignore_dufs,
2337         //                                                                          e_value_max,
2338         //                                                                          max_allowed_overlap,
2339         //                                                                          no_engulfing_overlaps,
2340         //                                                                          cutoff_scores_file );
2341         //            String s = "_";
2342         //            if ( radomize_fitch_parsimony ) {
2343         //                s += random_number_seed_for_fitch_parsimony + "_";
2344         //            }
2345         //            int i = 0;
2346         //            for( final Phylogeny inferred_tree : inferred_trees ) {
2347         //                if ( !inferred_tree.isRooted() ) {
2348         //                    intrees[ 0 ].getRoot().getName();
2349         //                    inferred_tree.r
2350         //                }
2351         //                final String outfile_name = ForesterUtil.removeSuffix( inferred_tree.getName() ) + s;
2352         //                final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator
2353         //                        .createInstance( inferred_tree, gwcd_list );
2354         //                SurfacingUtil.executeParsimonyAnalysis( random_number_seed_for_fitch_parsimony,
2355         //                                                        radomize_fitch_parsimony,
2356         //                                                        outfile_name,
2357         //                                                        domain_parsimony,
2358         //                                                        inferred_tree,
2359         //                                                        domain_id_to_go_ids_map,
2360         //                                                        go_id_to_term_map,
2361         //                                                        go_namespace_limit,
2362         //                                                        parameters_sb.toString() );
2363         //                i++;
2364         //            }
2365         //        }
2366         if ( all_bin_domain_combinations_gained_fitch != null ) {
2367             try {
2368                 executeFitchGainsAnalysis( new File( output_file
2369                                                    + surfacing.OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2370                                            all_bin_domain_combinations_gained_fitch,
2371                                            all_domains_encountered.size(),
2372                                            all_bin_domain_combinations_encountered,
2373                                            true );
2374             }
2375             catch ( final IOException e ) {
2376                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2377             }
2378         }
2379         if ( all_bin_domain_combinations_lost_fitch != null ) {
2380             try {
2381                 executeFitchGainsAnalysis( new File( output_file
2382                                                    + surfacing.OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2383                                            all_bin_domain_combinations_lost_fitch,
2384                                            all_domains_encountered.size(),
2385                                            all_bin_domain_combinations_encountered,
2386                                            false );
2387             }
2388             catch ( final IOException e ) {
2389                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2390             }
2391         }
2392         final Runtime rt = java.lang.Runtime.getRuntime();
2393         final long free_memory = rt.freeMemory() / 1000000;
2394         final long total_memory = rt.totalMemory() / 1000000;
2395         ForesterUtil.programMessage( PRG_NAME, "Time for analysis : " + ( new Date().getTime() - analysis_start_time )
2396                 + "ms" );
2397         ForesterUtil.programMessage( PRG_NAME, "Total running time: " + ( new Date().getTime() - start_time ) + "ms " );
2398         ForesterUtil.programMessage( PRG_NAME, "Free memory       : " + free_memory + "MB, total memory: "
2399                 + total_memory + "MB" );
2400         ForesterUtil.programMessage( PRG_NAME, "If this application is useful to you, please cite:" );
2401         ForesterUtil.programMessage( PRG_NAME, surfacing.WWW );
2402         ForesterUtil.programMessage( PRG_NAME, "OK" );
2403         System.out.println();
2404     }
2405
2406     private static void createSplitWriters( final File out_dir,
2407                                             final String my_outfile,
2408                                             final Map<Character, Writer> split_writers ) throws IOException {
2409         split_writers.put( 'a', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2410                 + "_domains_A.html" ) ) );
2411         split_writers.put( 'b', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2412                 + "_domains_B.html" ) ) );
2413         split_writers.put( 'c', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2414                 + "_domains_C.html" ) ) );
2415         split_writers.put( 'd', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2416                 + "_domains_D.html" ) ) );
2417         split_writers.put( 'e', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2418                 + "_domains_E.html" ) ) );
2419         split_writers.put( 'f', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2420                 + "_domains_F.html" ) ) );
2421         split_writers.put( 'g', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2422                 + "_domains_G.html" ) ) );
2423         split_writers.put( 'h', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2424                 + "_domains_H.html" ) ) );
2425         split_writers.put( 'i', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2426                 + "_domains_I.html" ) ) );
2427         split_writers.put( 'j', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2428                 + "_domains_J.html" ) ) );
2429         split_writers.put( 'k', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2430                 + "_domains_K.html" ) ) );
2431         split_writers.put( 'l', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2432                 + "_domains_L.html" ) ) );
2433         split_writers.put( 'm', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2434                 + "_domains_M.html" ) ) );
2435         split_writers.put( 'n', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2436                 + "_domains_N.html" ) ) );
2437         split_writers.put( 'o', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2438                 + "_domains_O.html" ) ) );
2439         split_writers.put( 'p', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2440                 + "_domains_P.html" ) ) );
2441         split_writers.put( 'q', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2442                 + "_domains_Q.html" ) ) );
2443         split_writers.put( 'r', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2444                 + "_domains_R.html" ) ) );
2445         split_writers.put( 's', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2446                 + "_domains_S.html" ) ) );
2447         split_writers.put( 't', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2448                 + "_domains_T.html" ) ) );
2449         split_writers.put( 'u', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2450                 + "_domains_U.html" ) ) );
2451         split_writers.put( 'v', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2452                 + "_domains_V.html" ) ) );
2453         split_writers.put( 'w', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2454                 + "_domains_W.html" ) ) );
2455         split_writers.put( 'x', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2456                 + "_domains_X.html" ) ) );
2457         split_writers.put( 'y', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2458                 + "_domains_Y.html" ) ) );
2459         split_writers.put( 'z', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2460                 + "_domains_Z.html" ) ) );
2461         split_writers.put( '0', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
2462                 + "_domains_0.html" ) ) );
2463     }
2464
2465     private static void printOutPercentageOfMultidomainProteins( final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
2466                                                                  final Writer log_writer ) {
2467         int sum = 0;
2468         for( final Entry<Integer, Integer> entry : all_genomes_domains_per_potein_histo.entrySet() ) {
2469             sum += entry.getValue();
2470         }
2471         final double percentage = 100.0 * ( sum - all_genomes_domains_per_potein_histo.get( 1 ) ) / sum;
2472         ForesterUtil.programMessage( PRG_NAME, "Percentage of multidomain proteins: " + percentage + "%" );
2473         log( "Percentage of multidomain proteins:            : " + percentage + "%", log_writer );
2474     }
2475
2476     private static void preparePhylogenyForParsimonyAnalyses( final Phylogeny intree,
2477                                                               final String[][] input_file_properties ) {
2478         final String[] genomes = new String[ input_file_properties.length ];
2479         for( int i = 0; i < input_file_properties.length; ++i ) {
2480             if ( intree.getNodes( input_file_properties[ i ][ 1 ] ).size() > 1 ) {
2481                 ForesterUtil.fatalError( surfacing.PRG_NAME, "node named [" + input_file_properties[ i ][ 1 ]
2482                         + "] is not unique in input tree " + intree.getName() );
2483             }
2484             genomes[ i ] = input_file_properties[ i ][ 1 ];
2485         }
2486         //
2487         final PhylogenyNodeIterator it = intree.iteratorPostorder();
2488         while ( it.hasNext() ) {
2489             final PhylogenyNode n = it.next();
2490             if ( ForesterUtil.isEmpty( n.getName() ) ) {
2491                 if ( n.getNodeData().isHasTaxonomy()
2492                         && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getTaxonomyCode() ) ) {
2493                     n.setName( n.getNodeData().getTaxonomy().getTaxonomyCode() );
2494                 }
2495                 else if ( n.getNodeData().isHasTaxonomy()
2496                         && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getScientificName() ) ) {
2497                     n.setName( n.getNodeData().getTaxonomy().getScientificName() );
2498                 }
2499                 else if ( n.getNodeData().isHasTaxonomy()
2500                         && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getCommonName() ) ) {
2501                     n.setName( n.getNodeData().getTaxonomy().getCommonName() );
2502                 }
2503                 else {
2504                     ForesterUtil
2505                             .fatalError( surfacing.PRG_NAME,
2506                                          "node with no name, scientific name, common name, or taxonomy code present" );
2507                 }
2508             }
2509         }
2510         //
2511         final List<String> igns = PhylogenyMethods.deleteExternalNodesPositiveSelection( genomes, intree );
2512         if ( igns.size() > 0 ) {
2513             System.out.println( "Not using the following " + igns.size() + " nodes:" );
2514             for( int i = 0; i < igns.size(); ++i ) {
2515                 System.out.println( " " + i + ": " + igns.get( i ) );
2516             }
2517             System.out.println( "--" );
2518         }
2519         for( int i = 0; i < input_file_properties.length; ++i ) {
2520             try {
2521                 intree.getNode( input_file_properties[ i ][ 1 ] );
2522             }
2523             catch ( final IllegalArgumentException e ) {
2524                 ForesterUtil.fatalError( surfacing.PRG_NAME, "node named [" + input_file_properties[ i ][ 1 ]
2525                         + "] not present/not unique in input tree" );
2526             }
2527         }
2528     }
2529
2530     // public static StringBuffer stringCombinableDomainsMapToStringBuffer(
2531     // final SortedMap<String, CombinableDomains> map ) {
2532     // final StringBuffer sb = new StringBuffer();
2533     // for( final Iterator<String> iter = map.keySet().iterator();
2534     // iter.hasNext(); ) {
2535     // final Object key = iter.next();
2536     // sb.append( ForesterUtil.pad( new StringBuffer( key.toString() ), 18, ' ',
2537     // false ) );
2538     // final CombinableDomains domain_combination = map.get( key );
2539     // sb.append( ForesterUtil.pad( new StringBuffer( "" +
2540     // domain_combination.getNumberOfCombiningDomains() ), 8,
2541     // ' ', false ) );
2542     // sb.append( domain_combination.toStringBuffer() );
2543     // sb.append( ForesterUtil.getLineSeparator() );
2544     // }
2545     // return sb;
2546     // }
2547     private static void printHelp() {
2548         System.out.println();
2549         System.out.println( "Usage:" );
2550         System.out.println();
2551         System.out.println( "% java -Xms256m -Xmx512m -cp forester.jar org.forester.applications." + surfacing.PRG_NAME
2552                 + " [options] <phylogen(y|ies) infile> [external node name 1] [name 2] ... [name n]" );
2553         System.out.println();
2554         System.out.println( " Note: This software might need a significant amount of memory (heap space);" );
2555         System.out
2556                 .println( "       hence use \"-Xms128m -Xmx512m\" (or more) to prevent a \"java.lang.OutOfMemoryError\"." );
2557         System.out.println();
2558         System.out.println( " Options: " );
2559         System.out.println( surfacing.DETAILEDNESS_OPTION + ": level of detail for similarities output file (default:"
2560                 + DETAILEDNESS_DEFAULT + ")" );
2561         System.out.println( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION
2562                 + ": to ignore combinations with self (default: not to ignore)" );
2563         System.out
2564                 .println( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION
2565                         + ": to ignore domains without combinations in any species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2566         System.out
2567                 .println( surfacing.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION
2568                         + ": to ignore domains specific to one species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2569         System.out.println( surfacing.NOT_IGNORE_DUFS_OPTION
2570                 + ": to _not_ ignore DUFs (domains with unknown function) (default: ignore DUFs)" );
2571         System.out
2572                 .println( surfacing.IGNORE_VIRAL_IDS
2573                         + ": to ignore domains with ids containing 'vir', 'retro', 'transpos', 'phage', or starting with 'rv' or 'gag_'" );
2574         System.out.println( surfacing.DOMAIN_SIMILARITY_SORT_OPTION + ": sorting for similarities (default: "
2575                 + DOMAIN_SORT_FILD_DEFAULT + ")" );
2576         System.out.println( surfacing.OUTPUT_FILE_OPTION + ": name for (main) output file (mandatory)" );
2577         System.out.println( surfacing.MAX_E_VALUE_OPTION + ": max (inclusive) E-value" );
2578         System.out.println( surfacing.MAX_ALLOWED_OVERLAP_OPTION + ": maximal allowed domain overlap" );
2579         System.out.println( surfacing.NO_ENGULFING_OVERLAP_OPTION + ": to ignore engulfed lower confidence domains" );
2580         System.out.println( surfacing.SPECIES_MATRIX_OPTION + ": species matrix" );
2581         System.out.println( surfacing.SCORING_OPTION + ": scoring (default:" + SCORING_DEFAULT + ")" );
2582         System.out.println( surfacing.DOMAIN_COUNT_SORT_OPTION + ": sorting for domain counts (default:"
2583                 + DOMAINS_SORT_ORDER_DEFAULT + ")" );
2584         System.out.println( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION + ": domain similarity print option (default:"
2585                 + DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT + ")" );
2586         System.out.println( surfacing.CUTOFF_SCORE_FILE_OPTION + ": cutoff score file" );
2587         System.out.println( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION
2588                 + ": sort by species count first" );
2589         System.out.println( surfacing.OUTPUT_DIR_OPTION + ": output directory" );
2590         System.out.println( surfacing.PFAM_TO_GO_FILE_USE_OPTION + ": Pfam to GO mapping file" );
2591         System.out.println( surfacing.GO_OBO_FILE_USE_OPTION + ": GO terms file (OBO format)" );
2592         System.out.println( surfacing.GO_NAMESPACE_LIMIT_OPTION + ": limit GO term to one GO namespace" );
2593         System.out.println( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
2594                 + "[=<suffix for pairwise comparison output files>]: to perform pairwise comparison based analyses" );
2595         System.out.println( surfacing.INPUT_SPECIES_TREE_OPTION
2596                 + ": species tree, to perform (Dollo, Fitch) parismony analyses" );
2597         System.out
2598                 .println( JACKNIFE_OPTION
2599                         + ": perform jacknife resampling for domain and binary domain combination based distance matrices [default resamplings: "
2600                         + JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT + "]" );
2601         System.out.println( JACKNIFE_RATIO_OPTION + ": ratio for jacknife resampling [default: "
2602                 + JACKNIFE_RATIO_DEFAULT + "]" );
2603         System.out.println( JACKNIFE_RANDOM_SEED_OPTION
2604                 + ": seed for random number generator for jacknife resampling [default: "
2605                 + JACKNIFE_RANDOM_SEED_DEFAULT + "]" );
2606         //        System.out.println( surfacing.INFER_SPECIES_TREES_OPTION
2607         //                + ": to infer NJ species trees based on shared domains/binary domain combinations" );
2608         System.out
2609                 .println( surfacing.INPUT_SPECIES_TREE_OPTION
2610                         + "=<treefiles in phyloXML format, separated by #>: to infer domain/binary domain combination gains/losses on given species trees" );
2611         System.out.println( surfacing.FILTER_POSITIVE_OPTION
2612                 + "=<file>: to filter out proteins not containing at least one domain listed in <file>" );
2613         System.out.println( surfacing.FILTER_NEGATIVE_OPTION
2614                 + "=<file>: to filter out proteins containing at least one domain listed in <file>" );
2615         System.out.println( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION
2616                 + "=<file>: to filter out (ignore) domains listed in <file>" );
2617         System.out.println( surfacing.INPUT_FILES_FROM_FILE_OPTION + "=<file>: to read input files from <file>" );
2618         System.out
2619                 .println( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION
2620                         + "=<seed>: seed for random number generator for Fitch Parsimony analysis (type: long, default: no randomization - given a choice, prefer absence" );
2621         System.out.println( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS
2622                 + ": to consider directedness in binary combinations: e.g. A-B != B-A" );
2623         System.out.println( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY
2624                 + ": to consider directedness and adjacency in binary combinations" );
2625         System.out
2626                 .println( surfacing.SEQ_EXTRACT_OPTION
2627                         + "=<domain ids (Pfam names)>: to extract sequence names of sequences containing matching domains and/or domain-sequences (order N to C) (domain separator: '~', domain sequences speparator: '#', e.g. 'NACHT#BIR~CARD')" );
2628         System.out.println( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE
2629                 + "=<file>: to perfom parsimony analysis on secondary features" );
2630         System.out.println( surfacing.PLUS_MINUS_ANALYSIS_OPTION + "=<file>: to presence/absence genome analysis" );
2631         System.out.println( surfacing.DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS
2632                 + ": to output binary domain combinations for (downstream) graph analysis" );
2633         System.out.println( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS + ": to output all proteins per domain" );
2634         System.out.println();
2635         System.out.println();
2636         System.out.println( "Example: java -Xms128m -Xmx512m -cp path/to/forester.jar"
2637                 + " org.forester.application.surfacing -detail=punctilious -o=TEST.html -pwc=TEST"
2638                 + " -cos=Pfam_ls_22_TC2 -p2g=pfam2go -obo=gene_ontology_edit.obo "
2639                 + "-dc_sort=dom -ignore_with_self -no_singles -e=0.001 -mo=1 -no_eo "
2640                 + "-ds_output=detailed_html -scoring=domains -sort=alpha -" + JACKNIFE_OPTION
2641                 + "=50 human mouse brafl strpu" );
2642         System.out.println();
2643     }
2644
2645     private static void processFilter( final File filter_file, final SortedSet<DomainId> filter ) {
2646         SortedSet<String> filter_str = null;
2647         try {
2648             filter_str = ForesterUtil.file2set( filter_file );
2649         }
2650         catch ( final IOException e ) {
2651             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2652         }
2653         if ( filter_str != null ) {
2654             for( final String string : filter_str ) {
2655                 filter.add( new DomainId( string ) );
2656             }
2657         }
2658         if ( VERBOSE ) {
2659             System.out.println( "Filter:" );
2660             for( final DomainId domainId : filter ) {
2661                 System.out.println( domainId.getId() );
2662             }
2663         }
2664     }
2665
2666     private static String[][] processInputFileNames( final String[] names ) {
2667         final String[][] input_file_properties = new String[ names.length ][];
2668         for( int i = 0; i < names.length; ++i ) {
2669             if ( names[ i ].indexOf( SEPARATOR_FOR_INPUT_VALUES ) < 0 ) {
2670                 input_file_properties[ i ] = new String[ 2 ];
2671                 input_file_properties[ i ][ 0 ] = names[ i ];
2672                 input_file_properties[ i ][ 1 ] = names[ i ];
2673             }
2674             else {
2675                 input_file_properties[ i ] = names[ i ].split( surfacing.SEPARATOR_FOR_INPUT_VALUES + "" );
2676                 if ( input_file_properties[ i ].length != 3 ) {
2677                     ForesterUtil
2678                             .fatalError( surfacing.PRG_NAME,
2679                                          "properties for the input files (hmmpfam output) are expected "
2680                                                  + "to be in the following format \"<hmmpfam output file>#<species>\" (or just one word, which is both the filename and the species id), instead received \""
2681                                                  + names[ i ] + "\"" );
2682                 }
2683             }
2684             final String error = ForesterUtil.isReadableFile( new File( input_file_properties[ i ][ 0 ] ) );
2685             if ( !ForesterUtil.isEmpty( error ) ) {
2686                 ForesterUtil.fatalError( surfacing.PRG_NAME, error );
2687             }
2688         }
2689         return input_file_properties;
2690     }
2691
2692     private static void processPlusMinusAnalysisOption( final CommandLineArguments cla,
2693                                                         final List<String> high_copy_base,
2694                                                         final List<String> high_copy_target,
2695                                                         final List<String> low_copy,
2696                                                         final List<Object> numbers ) {
2697         if ( cla.isOptionSet( surfacing.PLUS_MINUS_ANALYSIS_OPTION ) ) {
2698             if ( !cla.isOptionValueSet( surfacing.PLUS_MINUS_ANALYSIS_OPTION ) ) {
2699                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for 'plus-minus' file: -"
2700                         + surfacing.PLUS_MINUS_ANALYSIS_OPTION + "=<file>" );
2701             }
2702             final File plus_minus_file = new File( cla.getOptionValue( surfacing.PLUS_MINUS_ANALYSIS_OPTION ) );
2703             final String msg = ForesterUtil.isReadableFile( plus_minus_file );
2704             if ( !ForesterUtil.isEmpty( msg ) ) {
2705                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + plus_minus_file + "\": " + msg );
2706             }
2707             processPlusMinusFile( plus_minus_file, high_copy_base, high_copy_target, low_copy, numbers );
2708         }
2709     }
2710
2711     // First numbers is minimal difference, second is factor.
2712     private static void processPlusMinusFile( final File plus_minus_file,
2713                                               final List<String> high_copy_base,
2714                                               final List<String> high_copy_target,
2715                                               final List<String> low_copy,
2716                                               final List<Object> numbers ) {
2717         Set<String> species_set = null;
2718         int min_diff = PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT;
2719         double factor = PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT;
2720         try {
2721             species_set = ForesterUtil.file2set( plus_minus_file );
2722         }
2723         catch ( final IOException e ) {
2724             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2725         }
2726         if ( species_set != null ) {
2727             for( final String species : species_set ) {
2728                 final String species_trimmed = species.substring( 1 );
2729                 if ( species.startsWith( "+" ) ) {
2730                     if ( low_copy.contains( species_trimmed ) ) {
2731                         ForesterUtil.fatalError( surfacing.PRG_NAME,
2732                                                  "species/genome names can not appear with both '+' and '-' suffix, as appears the case for: \""
2733                                                          + species_trimmed + "\"" );
2734                     }
2735                     high_copy_base.add( species_trimmed );
2736                 }
2737                 else if ( species.startsWith( "*" ) ) {
2738                     if ( low_copy.contains( species_trimmed ) ) {
2739                         ForesterUtil.fatalError( surfacing.PRG_NAME,
2740                                                  "species/genome names can not appear with both '*' and '-' suffix, as appears the case for: \""
2741                                                          + species_trimmed + "\"" );
2742                     }
2743                     high_copy_target.add( species_trimmed );
2744                 }
2745                 else if ( species.startsWith( "-" ) ) {
2746                     if ( high_copy_base.contains( species_trimmed ) || high_copy_target.contains( species_trimmed ) ) {
2747                         ForesterUtil.fatalError( surfacing.PRG_NAME,
2748                                                  "species/genome names can not appear with both '+' or '*' and '-' suffix, as appears the case for: \""
2749                                                          + species_trimmed + "\"" );
2750                     }
2751                     low_copy.add( species_trimmed );
2752                 }
2753                 else if ( species.startsWith( "$D" ) ) {
2754                     try {
2755                         min_diff = Integer.parseInt( species.substring( 3 ) );
2756                     }
2757                     catch ( final NumberFormatException e ) {
2758                         ForesterUtil.fatalError( surfacing.PRG_NAME,
2759                                                  "could not parse integer value for minimal difference from: \""
2760                                                          + species.substring( 3 ) + "\"" );
2761                     }
2762                 }
2763                 else if ( species.startsWith( "$F" ) ) {
2764                     try {
2765                         factor = Double.parseDouble( species.substring( 3 ) );
2766                     }
2767                     catch ( final NumberFormatException e ) {
2768                         ForesterUtil.fatalError( surfacing.PRG_NAME, "could not parse double value for factor from: \""
2769                                 + species.substring( 3 ) + "\"" );
2770                     }
2771                 }
2772                 else if ( species.startsWith( "#" ) ) {
2773                     // Comment, ignore.
2774                 }
2775                 else {
2776                     ForesterUtil
2777                             .fatalError( surfacing.PRG_NAME,
2778                                          "species/genome names in 'plus minus' file must begin with '*' (high copy target genome), '+' (high copy base genomes), '-' (low copy genomes), '$D=<integer>' minimal Difference (default is 1), '$F=<double>' factor (default is 1.0), double), or '#' (ignore) suffix, encountered: \""
2779                                                  + species + "\"" );
2780                 }
2781                 numbers.add( new Integer( min_diff + "" ) );
2782                 numbers.add( new Double( factor + "" ) );
2783             }
2784         }
2785         else {
2786             ForesterUtil.fatalError( surfacing.PRG_NAME, "'plus minus' file [" + plus_minus_file + "] appears empty" );
2787         }
2788     }
2789
2790     private static void writePresentToNexus( final File output_file,
2791                                              final File positive_filter_file,
2792                                              final SortedSet<DomainId> filter,
2793                                              final List<GenomeWideCombinableDomains> gwcd_list ) {
2794         try {
2795             SurfacingUtil
2796                     .writeMatrixToFile( DomainParsimonyCalculator
2797                             .createMatrixOfDomainPresenceOrAbsence( gwcd_list, positive_filter_file == null ? null
2798                                     : filter ), output_file + DOMAINS_PRESENT_NEXUS, Format.NEXUS_BINARY );
2799             SurfacingUtil.writeMatrixToFile( DomainParsimonyCalculator
2800                     .createMatrixOfBinaryDomainCombinationPresenceOrAbsence( gwcd_list ), output_file
2801                     + BDC_PRESENT_NEXUS, Format.NEXUS_BINARY );
2802         }
2803         catch ( final Exception e ) {
2804             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
2805         }
2806     }
2807
2808     private static void writeProteinListsForAllSpecies( final File output_dir,
2809                                                         final SortedMap<Species, List<Protein>> protein_lists_per_species,
2810                                                         final List<GenomeWideCombinableDomains> gwcd_list ) {
2811         final SortedSet<DomainId> all_domains = new TreeSet<DomainId>();
2812         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
2813             all_domains.addAll( gwcd.getAllDomainIds() );
2814         }
2815         for( final DomainId domain : all_domains ) {
2816             final File out = new File( output_dir + ForesterUtil.FILE_SEPARATOR + domain + SEQ_EXTRACT_SUFFIX );
2817             SurfacingUtil.checkForOutputFileWriteability( out );
2818             try {
2819                 final Writer proteins_file_writer = new BufferedWriter( new FileWriter( out ) );
2820                 SurfacingUtil.extractProteinNames( protein_lists_per_species,
2821                                                    domain,
2822                                                    proteins_file_writer,
2823                                                    "\t",
2824                                                    LIMIT_SPEC_FOR_PROT_EX );
2825                 proteins_file_writer.close();
2826             }
2827             catch ( final IOException e ) {
2828                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
2829             }
2830             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote proteins list to \"" + out + "\"" );
2831         }
2832     }
2833 }