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: https://sites.google.com/site/cmzmasek/home/software/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.List;
  38 import java.util.Map;
  39 import java.util.Set;
  40 import java.util.SortedMap;
  41 import java.util.SortedSet;
  42 import java.util.TreeMap;
  43 import java.util.TreeSet;
  44
  45 import org.forester.go.GoId;
  46 import org.forester.go.GoNameSpace;
  47 import org.forester.go.GoTerm;
  48 import org.forester.go.GoUtils;
  49 import org.forester.go.OBOparser;
  50 import org.forester.go.PfamToGoMapping;
  51 import org.forester.go.PfamToGoParser;
  52 import org.forester.io.parsers.HmmscanPerDomainTableParser;
  53 import org.forester.io.parsers.HmmscanPerDomainTableParser.INDIVIDUAL_SCORE_CUTOFF;
  54 import org.forester.phylogeny.Phylogeny;
  55 import org.forester.protein.BinaryDomainCombination;
  56 import org.forester.protein.Domain;
  57 import org.forester.protein.Protein;
  58 import org.forester.species.BasicSpecies;
  59 import org.forester.species.Species;
  60 import org.forester.surfacing.BasicDomainSimilarityCalculator;
  61 import org.forester.surfacing.BasicGenomeWideCombinableDomains;
  62 import org.forester.surfacing.CombinationsBasedPairwiseDomainSimilarityCalculator;
  63 import org.forester.surfacing.DomainCountsBasedPairwiseSimilarityCalculator;
  64 import org.forester.surfacing.DomainLengthsTable;
  65 import org.forester.surfacing.DomainParsimonyCalculator;
  66 import org.forester.surfacing.DomainSimilarity;
  67 import org.forester.surfacing.DomainSimilarity.DomainSimilarityScoring;
  68 import org.forester.surfacing.DomainSimilarity.PRINT_OPTION;
  69 import org.forester.surfacing.DomainSimilarityCalculator;
  70 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  71 import org.forester.surfacing.GenomeWideCombinableDomains;
  72 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  73 import org.forester.surfacing.MappingResults;
  74 import org.forester.surfacing.PairwiseDomainSimilarityCalculator;
  75 import org.forester.surfacing.PairwiseGenomeComparator;
  76 import org.forester.surfacing.ProteinCountsBasedPairwiseDomainSimilarityCalculator;
  77 import org.forester.surfacing.SurfacingUtil;
  78 import org.forester.util.BasicDescriptiveStatistics;
  79 import org.forester.util.BasicTable;
  80 import org.forester.util.BasicTableParser;
  81 import org.forester.util.CommandLineArguments;
  82 import org.forester.util.DescriptiveStatistics;
  83 import org.forester.util.ForesterConstants;
  84 import org.forester.util.ForesterUtil;
  85
  86 public class surfacing {
  87
  88     private static final int                                        MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING                                  = 1000;
  89     public final static String                                      DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS                           = "graph_analysis_out";
  90     public final static String                                      DOMAIN_COMBINITONS_COUNTS_OUTPUT_OPTION                                       = "dcc";
  91     public final static String                                      DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS                       = "_dc.dot";
  92     public final static String                                      PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS        = "_fitch_present_dc.dot";
  93     public final static String                                      DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX                                    = ".dcc";
  94     // gain/loss:
  95     public final static String                                      PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS                                      = "_dollo_gl_d";
  96     public final static String                                      PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_BINARY_COMBINATIONS                          = "_dollo_gl_dc";
  97     public final static String                                      PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_DOMAINS                                      = "_fitch_gl_d";
  98     public final static String                                      PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS                          = "_fitch_gl_dc";
  99     // gain/loss counts:
 100     public final static String                                      PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS                               = "_dollo_glc_d";
 101     public final static String                                      PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_BINARY_COMBINATIONS                   = "_dollo_glc_dc";
 102     public final static String                                      PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_DOMAINS                               = "_fitch_glc_d";
 103     public final static String                                      PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS                   = "_fitch_glc_dc";
 104     // tables:
 105     public final static String                                      PARSIMONY_OUTPUT_FITCH_GAINS_BC                                               = "_fitch_gains_dc";
 106     public final static String                                      PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC                                          = "_fitch_gains_dc.html";
 107     public final static String                                      PARSIMONY_OUTPUT_FITCH_LOSSES_BC                                              = "_fitch_losses_dc";
 108     public final static String                                      PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC                                         = "_fitch_losses_dc.html";
 109     public final static String                                      PARSIMONY_OUTPUT_FITCH_PRESENT_BC                                             = "_fitch_present_dc";
 110     public final static String                                      PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC                                        = "_fitch_present_dc.html";
 111     public final static String                                      PARSIMONY_OUTPUT_DOLLO_GAINS_D                                                = "_dollo_gains_d";
 112     public final static String                                      PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D                                           = "_dollo_gains_d.html";
 113     public final static String                                      PARSIMONY_OUTPUT_DOLLO_LOSSES_D                                               = "_dollo_losses_d";
 114     public final static String                                      PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D                                          = "_dollo_losses_d.html";
 115     public final static String                                      PARSIMONY_OUTPUT_DOLLO_PRESENT_D                                              = "_dollo_present_d";
 116     public final static String                                      PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D                                         = "_dollo_present_d.html";
 117     public final static String                                      DOMAINS_PRESENT_NEXUS                                                         = "_dom.nex";
 118     public final static String                                      BDC_PRESENT_NEXUS                                                             = "_dc.nex";
 119     // ---
 120     public final static String                                      PRG_NAME                                                                      = "surfacing";
 121     public static final String                                      DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                                    = "_d_dollo"
 122             + ForesterConstants.PHYLO_XML_SUFFIX;
 123     public static final String                                      DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH                                    = "_d_fitch"
 124             + ForesterConstants.PHYLO_XML_SUFFIX;
 125     public static final String                                      BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                 = "_dc_dollo"
 126             + ForesterConstants.PHYLO_XML_SUFFIX;
 127     public static final String                                      BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH                 = "_dc_fitch"
 128             + ForesterConstants.PHYLO_XML_SUFFIX;
 129     public static final String                                      NEXUS_EXTERNAL_DOMAINS                                                        = "_dom.nex";
 130     public static final String                                      NEXUS_EXTERNAL_DOMAIN_COMBINATIONS                                            = "_dc.nex";
 131     public static final String                                      NEXUS_SECONDARY_FEATURES                                                      = "_secondary_features.nex";
 132     public static final String                                      PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES                           = "_dollo_gl_secondary_features";
 133     public static final String                                      PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES                    = "_dollo_glc_secondary_features";
 134     public static final String                                      PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES                               = "_dollo_gains_secondary_features";
 135     public static final String                                      PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES                              = "_dollo_losses_secondary_features";
 136     public static final String                                      PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES                             = "_dollo_present_secondary_features";
 137     public static final String                                      SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                         = "_secondary_features_dollo"
 138             + ForesterConstants.PHYLO_XML_SUFFIX;
 139     public static final String                                      PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_ALL_NAMESPACES                              = "_dollo_goid_d";
 140     public static final String                                      PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_ALL_NAMESPACES                             = "_fitch_goid_dc";
 141     final static private String                                     HELP_OPTION_1                                                                 = "help";
 142     final static private String                                     HELP_OPTION_2                                                                 = "h";
 143     final static private String                                     OUTPUT_DIR_OPTION                                                             = "out_dir";
 144     final static private String                                     SCORING_OPTION                                                                = "scoring";
 145     private static final DomainSimilarityScoring                    SCORING_DEFAULT                                                               = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 146     final static private String                                     SCORING_DOMAIN_COUNT_BASED                                                    = "domains";
 147     final static private String                                     SCORING_PROTEIN_COUNT_BASED                                                   = "proteins";
 148     final static private String                                     SCORING_COMBINATION_BASED                                                     = "combinations";
 149     final static private String                                     DETAILEDNESS_OPTION                                                           = "detail";
 150     private final static Detailedness                               DETAILEDNESS_DEFAULT                                                          = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 151     final static private String                                     SPECIES_MATRIX_OPTION                                                         = "smatrix";
 152     final static private String                                     DETAILEDNESS_BASIC                                                            = "basic";
 153     final static private String                                     DETAILEDNESS_LIST_IDS                                                         = "list_ids";
 154     final static private String                                     DETAILEDNESS_PUNCTILIOUS                                                      = "punctilious";
 155     final static private String                                     DOMAIN_SIMILARITY_SORT_OPTION                                                 = "sort";
 156     private static final DomainSimilarity.DomainSimilaritySortField DOMAIN_SORT_FILD_DEFAULT                                                      = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 157     final static private String                                     DOMAIN_SIMILARITY_SORT_MIN                                                    = "min";
 158     final static private String                                     DOMAIN_SIMILARITY_SORT_MAX                                                    = "max";
 159     final static private String                                     DOMAIN_SIMILARITY_SORT_SD                                                     = "sd";
 160     final static private String                                     DOMAIN_SIMILARITY_SORT_MEAN                                                   = "mean";
 161     final static private String                                     DOMAIN_SIMILARITY_SORT_DIFF                                                   = "diff";
 162     final static private String                                     DOMAIN_SIMILARITY_SORT_COUNTS_DIFF                                            = "count_diff";
 163     final static private String                                     DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF                                        = "abs_count_diff";
 164     final static private String                                     DOMAIN_SIMILARITY_SORT_SPECIES_COUNT                                          = "species";
 165     final static private String                                     DOMAIN_SIMILARITY_SORT_ALPHA                                                  = "alpha";
 166     final static private String                                     DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION                          = "species_first";
 167     final static private String                                     DOMAIN_COUNT_SORT_OPTION                                                      = "dc_sort";
 168     private static final GenomeWideCombinableDomainsSortOrder       DOMAINS_SORT_ORDER_DEFAULT                                                    = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
 169     final static private String                                     DOMAIN_COUNT_SORT_ALPHA                                                       = "alpha";
 170     final static private String                                     DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT                                            = "dom";
 171     final static private String                                     DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT                                   = "prot";
 172     final static private String                                     DOMAIN_COUNT_SORT_COMBINATIONS_COUNT                                          = "comb";
 173     final static private String                                     CUTOFF_SCORE_FILE_OPTION                                                      = "cos";
 174     final static private String                                     NOT_IGNORE_DUFS_OPTION                                                        = "dufs";
 175     final static private String                                     MAX_FS_E_VALUE_OPTION                                                         = "fs_e";
 176     final static private String                                     MAX_I_E_VALUE_OPTION                                                          = "ie";
 177     final static private String                                     MAX_ALLOWED_OVERLAP_OPTION                                                    = "mo";
 178     final static private String                                     NO_ENGULFING_OVERLAP_OPTION                                                   = "no_eo";
 179     final static private String                                     IGNORE_COMBINATION_WITH_SAME_OPTION                                           = "ignore_self_comb";
 180     final static private String                                     PERFORM_DC_REGAIN_PROTEINS_STATS_OPTION                                       = "dc_regain_stats";
 181     final static private String                                     DA_ANALYSIS_OPTION                                                            = "da_analyis";
 182     final static private String                                     USE_LAST_IN_FITCH_OPTION                                                      = "last";
 183     public final static String                                      PAIRWISE_DOMAIN_COMPARISONS_PREFIX                                            = "pwc_";
 184     final static private String                                     PAIRWISE_DOMAIN_COMPARISONS_OPTION                                            = "pwc";
 185     final static private String                                     OUTPUT_FILE_OPTION                                                            = "o";
 186     final static private String                                     PFAM_TO_GO_FILE_USE_OPTION                                                    = "p2g";
 187     final static private String                                     GO_OBO_FILE_USE_OPTION                                                        = "obo";
 188     final static private String                                     GO_NAMESPACE_LIMIT_OPTION                                                     = "go_namespace";
 189     final static private String                                     GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION                                  = "molecular_function";
 190     final static private String                                     GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS                                  = "biological_process";
 191     final static private String                                     GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT                                  = "cellular_component";
 192     final static private String                                     SECONDARY_FEATURES_PARSIMONY_MAP_FILE                                         = "secondary";
 193     final static private String                                     DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED                           = "simple_tab";
 194     final static private String                                     DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML                                    = "simple_html";
 195     final static private String                                     DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML                                  = "detailed_html";
 196     final static private String                                     DOMAIN_SIMILARITY_PRINT_OPTION                                                = "ds_output";
 197     private static final PRINT_OPTION                               DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT                                        = DomainSimilarity.PRINT_OPTION.HTML;
 198     final static private String                                     IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION                     = "ignore_singlet_domains";
 199     final static private String                                     IGNORE_VIRAL_IDS                                                              = "ignore_viral_ids";
 200     final static private boolean                                    IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT                    = false;
 201     final static private String                                     IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION                                 = "ignore_species_specific_domains";
 202     final static private boolean                                    IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT                         = false;
 203     final static private String                                     MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                                = "_mean_score.pwd";
 204     final static private String                                     MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                            = "_domains.pwd";
 205     final static private String                                     MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX                   = "_bin_combinations.pwd";
 206     final static private String                                     NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                               = "_mean_score_NJ"
 207             + ForesterConstants.PHYLO_XML_SUFFIX;
 208     final static private String                                     NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                           = "_domains_NJ"
 209             + ForesterConstants.PHYLO_XML_SUFFIX;
 210     final static private String                                     NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX                  = "_bin_combinations_NJ"
 211             + ForesterConstants.PHYLO_XML_SUFFIX;
 212     final static private String                                     FILTER_POSITIVE_OPTION                                                        = "pos_filter";
 213     final static private String                                     FILTER_NEGATIVE_OPTION                                                        = "neg_filter";
 214     final static private String                                     FILTER_NEGATIVE_DOMAINS_OPTION                                                = "neg_dom_filter";
 215     final static private String                                     INPUT_GENOMES_FILE_OPTION                                                     = "genomes";
 216     final static private String                                     INPUT_SPECIES_TREE_OPTION                                                     = "species_tree";
 217     final static private String                                     SEQ_EXTRACT_OPTION                                                            = "prot_extract";
 218     final static private String                                     PRG_VERSION                                                                   = "2.404";
 219     final static private String                                     PRG_DATE                                                                      = "170315";
 220     final static private String                                     E_MAIL                                                                        = "czmasek@burnham.org";
 221     final static private String                                     WWW                                                                           = "https://sites.google.com/site/cmzmasek/home/software/forester/surfacing";
 222     final static private boolean                                    IGNORE_DUFS_DEFAULT                                                           = true;
 223     final static private boolean                                    IGNORE_COMBINATION_WITH_SAME_DEFAULLT                                         = false;
 224     final static private double                                     MAX_E_VALUE_DEFAULT                                                           = -1;
 225     public final static int                                         MAX_ALLOWED_OVERLAP_DEFAULT                                                   = -1;
 226     private static final String                                     RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION                                        = "random_seed";
 227     private static final String                                     CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS                                      = "consider_bdc_direction";
 228     private static final String                                     CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY                        = "consider_bdc_adj";
 229     public static final String                                      SEQ_EXTRACT_SUFFIX                                                            = ".prot";
 230     public static final String                                      PLUS_MINUS_ANALYSIS_OPTION                                                    = "plus_minus";
 231     public static final String                                      PLUS_MINUS_DOM_SUFFIX                                                         = "_plus_minus_dom.txt";
 232     public static final String                                      PLUS_MINUS_DOM_SUFFIX_HTML                                                    = "_plus_minus_dom.html";
 233     public static final String                                      PLUS_MINUS_DC_SUFFIX_HTML                                                     = "_plus_minus_dc.html";
 234     public static final int                                         PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT                                          = 0;
 235     public static final double                                      PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT                                            = 1.0;
 236     public static final String                                      PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX                                              = "_plus_minus_go_ids_all.txt";
 237     public static final String                                      PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX                                          = "_plus_minus_go_ids_passing.txt";
 238     private static final String                                     OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS                                           = "all_prot";
 239     final static private String                                     OUTPUT_LIST_OF_ALL_PROTEINS_PER_DOMAIN_E_VALUE_OPTION                         = "all_prot_e";
 240     public static final boolean                                     VERBOSE                                                                       = false;
 241     private static final String                                     OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX              = "_fitch_dc_gains_counts";
 242     private static final String                                     OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX                = "_fitch_dc_losses_counts";
 243     private static final String                                     DOMAIN_LENGTHS_ANALYSIS_SUFFIX                                                = "_domain_lengths_analysis";
 244     private static final String                                     PERFORM_DOMAIN_LENGTH_ANALYSIS_OPTION                                         = "dla";
 245     public static final String                                      ALL_PFAMS_ENCOUNTERED_SUFFIX                                                  = "_all_encountered_pfams";
 246     public static final String                                      ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX                               = "_all_encountered_pfams_with_go_annotation";
 247     public static final String                                      ENCOUNTERED_PFAMS_SUMMARY_SUFFIX                                              = "_encountered_pfams_summary";
 248     public static final String                                      ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX                                            = "_all_pfams_gained_as_domains";
 249     public static final String                                      ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX                                              = "_all_pfams_lost_as_domains";
 250     public static final String                                      ALL_PFAMS_GAINED_AS_DC_SUFFIX                                                 = "_all_pfams_gained_as_dc";
 251     public static final String                                      ALL_PFAMS_LOST_AS_DC_SUFFIX                                                   = "_all_pfams_lost_as_dc";
 252     public static final String                                      BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES                                = "PER_NODE_EVENTS";
 253     public static final String                                      BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES                             = "PER_SUBTREE_EVENTS";
 254     public static final String                                      D_PROMISCUITY_FILE_SUFFIX                                                     = "_domain_promiscuities";
 255     private static final String                                     LOG_FILE_SUFFIX                                                               = "_log.txt";
 256     private static final String                                     DATA_FILE_SUFFIX                                                              = "_domain_combination_data.txt";
 257     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";
 258     private static final String                                     WRITE_TO_NEXUS_OPTION                                                         = "nexus";
 259     private static final String                                     PERFORM_DC_FITCH                                                              = "dc_pars";
 260     private static final INDIVIDUAL_SCORE_CUTOFF                    INDIVIDUAL_SCORE_CUTOFF_DEFAULT                                               = INDIVIDUAL_SCORE_CUTOFF.FULL_SEQUENCE;                                                                                                                                                      //TODO look at me! change?
 261     public static final String                                      INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX                          = "_indep_dc_gains_fitch_counts.txt";
 262     public static final String                                      INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX                              = "_indep_dc_gains_fitch_lists.txt";
 263     public static final String                                      INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX               = "_indep_dc_gains_fitch_lists_for_go_mapping.txt";
 264     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";
 265     public static final String                                      LIMIT_SPEC_FOR_PROT_EX                                                        = null;                                                                                                                                                                                       // e.g. "HUMAN"; set to null for not using this feature (default).
 266     public static final String                                      BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH_MAPPED          = "_dc_MAPPED_secondary_features_fitch"
 267             + ForesterConstants.PHYLO_XML_SUFFIX;
 268     public static final String                                      INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_MAPPED_OUTPUT_SUFFIX                   = "_indep_dc_gains_fitch_counts_MAPPED.txt";
 269     public static final String                                      INDEPENDENT_DC_GAINS_FITCH_PARS_DC_MAPPED_OUTPUT_SUFFIX                       = "_indep_dc_gains_fitch_lists_MAPPED.txt";
 270     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";
 271     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";
 272     private static final boolean                                    CALC_SIMILARITY_SCORES                                                        = false;
 273     @SuppressWarnings("unchecked")
 274     public static void main( final String args[] ) {
 275         final long start_time = new Date().getTime();
 276         // final StringBuffer log = new StringBuffer();
 277         final StringBuilder html_desc = new StringBuilder();
 278         ForesterUtil.printProgramInformation( surfacing.PRG_NAME,
 279                                               surfacing.PRG_VERSION,
 280                                               surfacing.PRG_DATE,
 281                                               surfacing.E_MAIL,
 282                                               surfacing.WWW );
 283         final String nl = ForesterUtil.LINE_SEPARATOR;
 284         html_desc.append( "<table>" + nl );
 285         html_desc.append( "<tr><td>Produced by:</td><td>" + surfacing.PRG_NAME + "</td></tr>" + nl );
 286         html_desc.append( "<tr><td>Version:</td><td>" + surfacing.PRG_VERSION + "</td></tr>" + nl );
 287         html_desc.append( "<tr><td>Release Date:</td><td>" + surfacing.PRG_DATE + "</td></tr>" + nl );
 288         html_desc.append( "<tr><td>Contact:</td><td>" + surfacing.E_MAIL + "</td></tr>" + nl );
 289         html_desc.append( "<tr><td>WWW:</td><td>" + surfacing.WWW + "</td></tr>" + nl );
 290         CommandLineArguments cla = null;
 291         try {
 292             cla = new CommandLineArguments( args );
 293         }
 294         catch ( final Exception e ) {
 295             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
 296         }
 297         if ( cla.isOptionSet( surfacing.HELP_OPTION_1 ) || cla.isOptionSet( surfacing.HELP_OPTION_2 ) ) {
 298             surfacing.printHelp();
 299             System.exit( 0 );
 300         }
 301         if ( ( args.length < 1 ) ) {
 302             surfacing.printHelp();
 303             System.exit( -1 );
 304         }
 305         final List<String> allowed_options = new ArrayList<String>();
 306         allowed_options.add( surfacing.NOT_IGNORE_DUFS_OPTION );
 307         allowed_options.add( surfacing.MAX_FS_E_VALUE_OPTION );
 308         allowed_options.add( surfacing.MAX_I_E_VALUE_OPTION );
 309         allowed_options.add( surfacing.DETAILEDNESS_OPTION );
 310         allowed_options.add( surfacing.OUTPUT_FILE_OPTION );
 311         allowed_options.add( surfacing.DOMAIN_SIMILARITY_SORT_OPTION );
 312         allowed_options.add( surfacing.SPECIES_MATRIX_OPTION );
 313         allowed_options.add( surfacing.SCORING_OPTION );
 314         allowed_options.add( surfacing.MAX_ALLOWED_OVERLAP_OPTION );
 315         allowed_options.add( surfacing.NO_ENGULFING_OVERLAP_OPTION );
 316         allowed_options.add( surfacing.DOMAIN_COUNT_SORT_OPTION );
 317         allowed_options.add( surfacing.CUTOFF_SCORE_FILE_OPTION );
 318         allowed_options.add( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION );
 319         allowed_options.add( surfacing.OUTPUT_DIR_OPTION );
 320         allowed_options.add( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION );
 321         allowed_options.add( surfacing.PFAM_TO_GO_FILE_USE_OPTION );
 322         allowed_options.add( surfacing.GO_OBO_FILE_USE_OPTION );
 323         allowed_options.add( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION );
 324         allowed_options.add( surfacing.GO_NAMESPACE_LIMIT_OPTION );
 325         allowed_options.add( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 326         allowed_options.add( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION );
 327         allowed_options.add( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS );
 328         allowed_options.add( INPUT_SPECIES_TREE_OPTION );
 329         allowed_options.add( FILTER_POSITIVE_OPTION );
 330         allowed_options.add( FILTER_NEGATIVE_OPTION );
 331         allowed_options.add( INPUT_GENOMES_FILE_OPTION );
 332         allowed_options.add( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
 333         allowed_options.add( FILTER_NEGATIVE_DOMAINS_OPTION );
 334         allowed_options.add( IGNORE_VIRAL_IDS );
 335         allowed_options.add( SEQ_EXTRACT_OPTION );
 336         allowed_options.add( OUTPUT_LIST_OF_ALL_PROTEINS_PER_DOMAIN_E_VALUE_OPTION );
 337         allowed_options.add( SECONDARY_FEATURES_PARSIMONY_MAP_FILE );
 338         allowed_options.add( PLUS_MINUS_ANALYSIS_OPTION );
 339         allowed_options.add( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS );
 340         allowed_options.add( DOMAIN_COMBINITONS_COUNTS_OUTPUT_OPTION );
 341         allowed_options.add( OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS );
 342         allowed_options.add( CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY );
 343         allowed_options.add( WRITE_TO_NEXUS_OPTION );
 344         allowed_options.add( PERFORM_DC_REGAIN_PROTEINS_STATS_OPTION );
 345         allowed_options.add( DA_ANALYSIS_OPTION );
 346         allowed_options.add( USE_LAST_IN_FITCH_OPTION );
 347         allowed_options.add( PERFORM_DC_FITCH );
 348         allowed_options.add( PERFORM_DOMAIN_LENGTH_ANALYSIS_OPTION );
 349         boolean ignore_dufs = surfacing.IGNORE_DUFS_DEFAULT;
 350         boolean ignore_combination_with_same = surfacing.IGNORE_COMBINATION_WITH_SAME_DEFAULLT;
 351         double fs_e_value_max = surfacing.MAX_E_VALUE_DEFAULT;
 352         double ie_value_max = surfacing.MAX_E_VALUE_DEFAULT;
 353         int max_allowed_overlap = surfacing.MAX_ALLOWED_OVERLAP_DEFAULT;
 354         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 355         if ( dissallowed_options.length() > 0 ) {
 356             ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown option(s): " + dissallowed_options );
 357         }
 358         boolean use_last_in_fitch_parsimony = false;
 359         if ( cla.isOptionSet( USE_LAST_IN_FITCH_OPTION ) ) {
 360             use_last_in_fitch_parsimony = true;
 361         }
 362         boolean write_to_nexus = false;
 363         if ( cla.isOptionSet( WRITE_TO_NEXUS_OPTION ) ) {
 364             write_to_nexus = true;
 365         }
 366         boolean perform_dc_fich = false;
 367         if ( cla.isOptionSet( PERFORM_DC_FITCH ) ) {
 368             perform_dc_fich = true;
 369         }
 370         boolean perform_dc_regain_proteins_stats = false;
 371         if ( cla.isOptionSet( PERFORM_DC_REGAIN_PROTEINS_STATS_OPTION ) ) {
 372             perform_dc_regain_proteins_stats = true;
 373         }
 374         boolean da_analysis = false;
 375         if ( cla.isOptionSet( DA_ANALYSIS_OPTION ) ) {
 376             da_analysis = true;
 377         }
 378         boolean output_binary_domain_combinationsfor_graph_analysis = false;
 379         if ( cla.isOptionSet( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS ) ) {
 380             output_binary_domain_combinationsfor_graph_analysis = true;
 381         }
 382         boolean output_binary_domain_combinationsfor_counts = false;
 383         if ( cla.isOptionSet( DOMAIN_COMBINITONS_COUNTS_OUTPUT_OPTION ) ) {
 384             output_binary_domain_combinationsfor_counts = true;
 385         }
 386         if ( cla.isOptionSet( surfacing.MAX_FS_E_VALUE_OPTION ) ) {
 387             try {
 388                 fs_e_value_max = cla.getOptionValueAsDouble( surfacing.MAX_FS_E_VALUE_OPTION );
 389             }
 390             catch ( final Exception e ) {
 391                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for E-value maximum" );
 392             }
 393         }
 394         if ( cla.isOptionSet( surfacing.MAX_I_E_VALUE_OPTION ) ) {
 395             try {
 396                 ie_value_max = cla.getOptionValueAsDouble( surfacing.MAX_I_E_VALUE_OPTION );
 397             }
 398             catch ( final Exception e ) {
 399                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for E-value maximum" );
 400             }
 401         }
 402         if ( cla.isOptionSet( surfacing.MAX_ALLOWED_OVERLAP_OPTION ) ) {
 403             try {
 404                 max_allowed_overlap = cla.getOptionValueAsInt( surfacing.MAX_ALLOWED_OVERLAP_OPTION );
 405             }
 406             catch ( final Exception e ) {
 407                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for maximal allowed domain overlap" );
 408             }
 409         }
 410         boolean no_engulfing_overlaps = false;
 411         if ( cla.isOptionSet( surfacing.NO_ENGULFING_OVERLAP_OPTION ) ) {
 412             no_engulfing_overlaps = true;
 413         }
 414         boolean ignore_virus_like_ids = false;
 415         if ( cla.isOptionSet( surfacing.IGNORE_VIRAL_IDS ) ) {
 416             ignore_virus_like_ids = true;
 417         }
 418         if ( cla.isOptionSet( surfacing.NOT_IGNORE_DUFS_OPTION ) ) {
 419             ignore_dufs = false;
 420         }
 421         if ( cla.isOptionSet( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION ) ) {
 422             ignore_combination_with_same = true;
 423         }
 424         boolean domain_length_analysis = false;
 425         if ( cla.isOptionSet( surfacing.PERFORM_DOMAIN_LENGTH_ANALYSIS_OPTION ) ) {
 426             domain_length_analysis = true;
 427         }
 428         boolean ignore_domains_without_combs_in_all_spec = IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT;
 429         if ( cla.isOptionSet( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION ) ) {
 430             ignore_domains_without_combs_in_all_spec = true;
 431         }
 432         boolean ignore_species_specific_domains = IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT;
 433         if ( cla.isOptionSet( surfacing.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION ) ) {
 434             ignore_species_specific_domains = true;
 435         }
 436         if ( !cla.isOptionValueSet( surfacing.INPUT_SPECIES_TREE_OPTION ) ) {
 437             ForesterUtil.fatalError( surfacing.PRG_NAME, "no input species tree file given: "
 438                     + surfacing.INPUT_SPECIES_TREE_OPTION + "=<file>" );
 439         }
 440         File output_file = null;
 441         if ( cla.isOptionSet( surfacing.OUTPUT_FILE_OPTION ) ) {
 442             if ( !cla.isOptionValueSet( surfacing.OUTPUT_FILE_OPTION ) ) {
 443                 ForesterUtil.fatalError( surfacing.PRG_NAME,
 444                                          "no value for domain combinations similarities output file: -"
 445                                                  + surfacing.OUTPUT_FILE_OPTION + "=<file>" );
 446             }
 447             output_file = new File( cla.getOptionValue( surfacing.OUTPUT_FILE_OPTION ) );
 448             SurfacingUtil.checkForOutputFileWriteability( output_file );
 449         }
 450         File cutoff_scores_file = null;
 451         Map<String, Double> individual_score_cutoffs = null;
 452         if ( cla.isOptionSet( surfacing.CUTOFF_SCORE_FILE_OPTION ) ) {
 453             if ( !cla.isOptionValueSet( surfacing.CUTOFF_SCORE_FILE_OPTION ) ) {
 454                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for individual domain score cutoffs file: -"
 455                         + surfacing.CUTOFF_SCORE_FILE_OPTION + "=<file>" );
 456             }
 457             cutoff_scores_file = new File( cla.getOptionValue( surfacing.CUTOFF_SCORE_FILE_OPTION ) );
 458             final String error = ForesterUtil.isReadableFile( cutoff_scores_file );
 459             if ( !ForesterUtil.isEmpty( error ) ) {
 460                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read individual domain score cutoffs file: "
 461                         + error );
 462             }
 463             try {
 464                 final BasicTable<String> scores_table = BasicTableParser.parse( cutoff_scores_file, ' ' );
 465                 individual_score_cutoffs = scores_table.getColumnsAsMapDouble( 0, 1 );
 466             }
 467             catch ( final IOException e ) {
 468                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from individual score cutoffs file: " + e );
 469             }
 470         }
 471         BinaryDomainCombination.DomainCombinationType dc_type = BinaryDomainCombination.DomainCombinationType.BASIC;
 472         if ( cla.isOptionSet( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS ) ) {
 473             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED;
 474         }
 475         if ( cla.isOptionSet( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY ) ) {
 476             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT;
 477         }
 478         File out_dir = null;
 479         if ( cla.isOptionSet( surfacing.OUTPUT_DIR_OPTION ) ) {
 480             if ( !cla.isOptionValueSet( surfacing.OUTPUT_DIR_OPTION ) ) {
 481                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for output directory: -"
 482                         + surfacing.OUTPUT_DIR_OPTION + "=<dir>" );
 483             }
 484             out_dir = new File( cla.getOptionValue( surfacing.OUTPUT_DIR_OPTION ) );
 485             if ( out_dir.exists() && ( out_dir.listFiles().length > 0 ) ) {
 486                 ForesterUtil.fatalError( surfacing.PRG_NAME, "\"" + out_dir + "\" aready exists and is not empty" );
 487             }
 488             if ( !out_dir.exists() ) {
 489                 final boolean success = out_dir.mkdir();
 490                 if ( !success || !out_dir.exists() ) {
 491                     ForesterUtil.fatalError( surfacing.PRG_NAME, "failed to create \"" + out_dir + "\"" );
 492                 }
 493             }
 494             if ( !out_dir.canWrite() ) {
 495                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot write to \"" + out_dir + "\"" );
 496             }
 497         }
 498         File positive_filter_file = null;
 499         File negative_filter_file = null;
 500         File negative_domains_filter_file = null;
 501         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) && cla.isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 502             ForesterUtil.fatalError( surfacing.PRG_NAME, "attempt to use both negative and positive protein filter" );
 503         }
 504         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION )
 505                 && ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) || cla
 506                         .isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) ) {
 507             ForesterUtil
 508             .fatalError( surfacing.PRG_NAME,
 509                          "attempt to use both negative or positive protein filter together wirh a negative domains filter" );
 510         }
 511         if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_OPTION ) ) {
 512             if ( !cla.isOptionValueSet( surfacing.FILTER_NEGATIVE_OPTION ) ) {
 513                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for negative filter: -"
 514                         + surfacing.FILTER_NEGATIVE_OPTION + "=<file>" );
 515             }
 516             negative_filter_file = new File( cla.getOptionValue( surfacing.FILTER_NEGATIVE_OPTION ) );
 517             final String msg = ForesterUtil.isReadableFile( negative_filter_file );
 518             if ( !ForesterUtil.isEmpty( msg ) ) {
 519                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + negative_filter_file + "\": "
 520                         + msg );
 521             }
 522         }
 523         else if ( cla.isOptionSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 524             if ( !cla.isOptionValueSet( surfacing.FILTER_POSITIVE_OPTION ) ) {
 525                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for positive filter: -"
 526                         + surfacing.FILTER_POSITIVE_OPTION + "=<file>" );
 527             }
 528             positive_filter_file = new File( cla.getOptionValue( surfacing.FILTER_POSITIVE_OPTION ) );
 529             final String msg = ForesterUtil.isReadableFile( positive_filter_file );
 530             if ( !ForesterUtil.isEmpty( msg ) ) {
 531                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + positive_filter_file + "\": "
 532                         + msg );
 533             }
 534         }
 535         else if ( cla.isOptionSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 536             if ( !cla.isOptionValueSet( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 537                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for negative domains filter: -"
 538                         + surfacing.FILTER_NEGATIVE_DOMAINS_OPTION + "=<file>" );
 539             }
 540             negative_domains_filter_file = new File( cla.getOptionValue( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION ) );
 541             final String msg = ForesterUtil.isReadableFile( negative_domains_filter_file );
 542             if ( !ForesterUtil.isEmpty( msg ) ) {
 543                 ForesterUtil.fatalError( surfacing.PRG_NAME, "can not read from \"" + negative_domains_filter_file
 544                                          + "\": " + msg );
 545             }
 546         }
 547         final List<String> plus_minus_analysis_high_copy_base_species = new ArrayList<String>();
 548         final List<String> plus_minus_analysis_high_copy_target_species = new ArrayList<String>();
 549         final List<String> plus_minus_analysis_high_low_copy_species = new ArrayList<String>();
 550         final List<Object> plus_minus_analysis_numbers = new ArrayList<Object>();
 551         SurfacingUtil.processPlusMinusAnalysisOption( cla,
 552                                                       plus_minus_analysis_high_copy_base_species,
 553                                                       plus_minus_analysis_high_copy_target_species,
 554                                                       plus_minus_analysis_high_low_copy_species,
 555                                                       plus_minus_analysis_numbers );
 556         File input_genomes_file = null;
 557         if ( cla.isOptionSet( surfacing.INPUT_GENOMES_FILE_OPTION ) ) {
 558             if ( !cla.isOptionValueSet( surfacing.INPUT_GENOMES_FILE_OPTION ) ) {
 559                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for input genomes file: -"
 560                         + surfacing.INPUT_GENOMES_FILE_OPTION + "=<file>" );
 561             }
 562             input_genomes_file = new File( cla.getOptionValue( surfacing.INPUT_GENOMES_FILE_OPTION ) );
 563             final String msg = ForesterUtil.isReadableFile( input_genomes_file );
 564             if ( !ForesterUtil.isEmpty( msg ) ) {
 565                 ForesterUtil
 566                 .fatalError( surfacing.PRG_NAME, "can not read from \"" + input_genomes_file + "\": " + msg );
 567             }
 568         }
 569         else {
 570             ForesterUtil.fatalError( surfacing.PRG_NAME, "no input genomes file given: "
 571                     + surfacing.INPUT_GENOMES_FILE_OPTION + "=<file>" );
 572         }
 573         DomainSimilarity.DomainSimilarityScoring scoring = SCORING_DEFAULT;
 574         if ( cla.isOptionSet( surfacing.SCORING_OPTION ) ) {
 575             if ( !cla.isOptionValueSet( surfacing.SCORING_OPTION ) ) {
 576                 ForesterUtil.fatalError( surfacing.PRG_NAME,
 577                                          "no value for scoring method for domain combinations similarity calculation: -"
 578                                                  + surfacing.SCORING_OPTION + "=<"
 579                                                  + surfacing.SCORING_DOMAIN_COUNT_BASED + "|"
 580                                                  + surfacing.SCORING_PROTEIN_COUNT_BASED + "|"
 581                                                  + surfacing.SCORING_COMBINATION_BASED + ">\"" );
 582             }
 583             final String scoring_str = cla.getOptionValue( surfacing.SCORING_OPTION );
 584             if ( scoring_str.equals( surfacing.SCORING_DOMAIN_COUNT_BASED ) ) {
 585                 scoring = DomainSimilarity.DomainSimilarityScoring.DOMAINS;
 586             }
 587             else if ( scoring_str.equals( surfacing.SCORING_COMBINATION_BASED ) ) {
 588                 scoring = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 589             }
 590             else if ( scoring_str.equals( surfacing.SCORING_PROTEIN_COUNT_BASED ) ) {
 591                 scoring = DomainSimilarity.DomainSimilarityScoring.PROTEINS;
 592             }
 593             else {
 594                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + scoring_str
 595                                          + "\" for scoring method for domain combinations similarity calculation: \"-"
 596                                          + surfacing.SCORING_OPTION + "=<" + surfacing.SCORING_DOMAIN_COUNT_BASED + "|"
 597                                          + surfacing.SCORING_PROTEIN_COUNT_BASED + "|" + surfacing.SCORING_COMBINATION_BASED + ">\"" );
 598             }
 599         }
 600         boolean sort_by_species_count_first = false;
 601         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION ) ) {
 602             sort_by_species_count_first = true;
 603         }
 604         boolean species_matrix = false;
 605         if ( cla.isOptionSet( surfacing.SPECIES_MATRIX_OPTION ) ) {
 606             species_matrix = true;
 607         }
 608         boolean output_protein_lists_for_all_domains = false;
 609         double output_list_of_all_proteins_per_domain_e_value_max = -1;
 610         if ( cla.isOptionSet( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS ) ) {
 611             output_protein_lists_for_all_domains = true;
 612             if ( cla.isOptionSet( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_PER_DOMAIN_E_VALUE_OPTION ) ) {
 613                 try {
 614                     output_list_of_all_proteins_per_domain_e_value_max = cla
 615                             .getOptionValueAsDouble( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_PER_DOMAIN_E_VALUE_OPTION );
 616                 }
 617                 catch ( final Exception e ) {
 618                     ForesterUtil.fatalError( surfacing.PRG_NAME, "no acceptable value for per domain E-value maximum" );
 619                 }
 620             }
 621         }
 622         Detailedness detailedness = DETAILEDNESS_DEFAULT;
 623         if ( cla.isOptionSet( surfacing.DETAILEDNESS_OPTION ) ) {
 624             if ( !cla.isOptionValueSet( surfacing.DETAILEDNESS_OPTION ) ) {
 625                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for -" + surfacing.DETAILEDNESS_OPTION + "=<"
 626                         + surfacing.DETAILEDNESS_BASIC + "|" + surfacing.DETAILEDNESS_LIST_IDS + "|"
 627                         + surfacing.DETAILEDNESS_PUNCTILIOUS + ">\"" );
 628             }
 629             final String detness = cla.getOptionValue( surfacing.DETAILEDNESS_OPTION ).toLowerCase();
 630             if ( detness.equals( surfacing.DETAILEDNESS_BASIC ) ) {
 631                 detailedness = DomainSimilarityCalculator.Detailedness.BASIC;
 632             }
 633             else if ( detness.equals( surfacing.DETAILEDNESS_LIST_IDS ) ) {
 634                 detailedness = DomainSimilarityCalculator.Detailedness.LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES;
 635             }
 636             else if ( detness.equals( surfacing.DETAILEDNESS_PUNCTILIOUS ) ) {
 637                 detailedness = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 638             }
 639             else {
 640                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + detness + "\" for detailedness: \"-"
 641                         + surfacing.DETAILEDNESS_OPTION + "=<" + surfacing.DETAILEDNESS_BASIC + "|"
 642                         + surfacing.DETAILEDNESS_LIST_IDS + "|" + surfacing.DETAILEDNESS_PUNCTILIOUS + ">\"" );
 643             }
 644         }
 645         String automated_pairwise_comparison_suffix = null;
 646         boolean perform_pwc = false;
 647         boolean write_pwc_files = false;
 648         if ( cla.isOptionSet( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 649             perform_pwc = true;
 650             if ( !cla.isOptionValueSet( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 651                 write_pwc_files = false;
 652             }
 653             else {
 654                 write_pwc_files = true;
 655                 automated_pairwise_comparison_suffix = "_"
 656                         + cla.getOptionValue( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 657             }
 658         }
 659         String query_domain_ids = null;
 660         if ( cla.isOptionSet( surfacing.SEQ_EXTRACT_OPTION ) ) {
 661             if ( !cla.isOptionValueSet( surfacing.SEQ_EXTRACT_OPTION ) ) {
 662                 ForesterUtil
 663                 .fatalError( surfacing.PRG_NAME,
 664                              "no domain ids given for sequences with given domains to be extracted : -"
 665                                      + surfacing.SEQ_EXTRACT_OPTION
 666                                      + "=<ordered domain sequences, domain ids separated by '~', sequences separated by '#'>" );
 667             }
 668             query_domain_ids = cla.getOptionValue( surfacing.SEQ_EXTRACT_OPTION );
 669         }
 670         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field = DOMAIN_SORT_FILD_DEFAULT;
 671         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field_for_automated_pwc = DOMAIN_SORT_FILD_DEFAULT;
 672         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 673             if ( !cla.isOptionValueSet( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 674                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for domain combinations similarities sorting: -"
 675                         + surfacing.DOMAIN_SIMILARITY_SORT_OPTION + "=<" + surfacing.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 676                         + surfacing.DOMAIN_SIMILARITY_SORT_MAX + "|" + surfacing.DOMAIN_SIMILARITY_SORT_MIN + "|"
 677                         + surfacing.DOMAIN_SIMILARITY_SORT_MEAN + "|" + surfacing.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 678                         + surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 679                         + surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|"
 680                         + surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|" + surfacing.DOMAIN_SIMILARITY_SORT_SD
 681                         + ">\"" );
 682             }
 683             final String sort_str = cla.getOptionValue( surfacing.DOMAIN_SIMILARITY_SORT_OPTION ).toLowerCase();
 684             if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_ALPHA ) ) {
 685                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 686                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 687             }
 688             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MAX ) ) {
 689                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX;
 690                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 691             }
 692             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MIN ) ) {
 693                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MIN;
 694                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 695             }
 696             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_MEAN ) ) {
 697                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MEAN;
 698                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MEAN;
 699             }
 700             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT ) ) {
 701                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SPECIES_COUNT;
 702                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 703             }
 704             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_SD ) ) {
 705                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SD;
 706                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 707             }
 708             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_DIFF ) ) {
 709                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 710                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 711             }
 712             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF ) ) {
 713                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 714                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 715             }
 716             else if ( sort_str.equals( surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF ) ) {
 717                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 718                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 719             }
 720             else {
 721                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort_str
 722                                          + "\" for domain combinations similarities sorting: \"-"
 723                                          + surfacing.DOMAIN_SIMILARITY_SORT_OPTION + "=<" + surfacing.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 724                                          + surfacing.DOMAIN_SIMILARITY_SORT_MAX + "|" + surfacing.DOMAIN_SIMILARITY_SORT_MIN + "|"
 725                                          + surfacing.DOMAIN_SIMILARITY_SORT_MEAN + "|" + surfacing.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 726                                          + surfacing.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 727                                          + surfacing.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|" + "|"
 728                                          + surfacing.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|" + surfacing.DOMAIN_SIMILARITY_SORT_SD
 729                                          + ">\"" );
 730             }
 731         }
 732         DomainSimilarity.PRINT_OPTION domain_similarity_print_option = DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT;
 733         if ( cla.isOptionSet( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 734             if ( !cla.isOptionValueSet( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 735                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for print option: -"
 736                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML + "|"
 737                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
 738                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
 739             }
 740             final String sort = cla.getOptionValue( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION ).toLowerCase();
 741             if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML ) ) {
 742                 domain_similarity_print_option = DomainSimilarity.PRINT_OPTION.HTML;
 743             }
 744             else if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML ) ) {
 745                 ForesterUtil.fatalError( surfacing.PRG_NAME, "simple HTML output not implemented yet :(" );
 746             }
 747             else if ( sort.equals( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED ) ) {
 748                 domain_similarity_print_option = DomainSimilarity.PRINT_OPTION.SIMPLE_TAB_DELIMITED;
 749             }
 750             else {
 751                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort + "\" for print option: -"
 752                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML + "|"
 753                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
 754                         + surfacing.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
 755             }
 756         }
 757         GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order = DOMAINS_SORT_ORDER_DEFAULT;
 758         if ( cla.isOptionSet( surfacing.DOMAIN_COUNT_SORT_OPTION ) ) {
 759             if ( !cla.isOptionValueSet( surfacing.DOMAIN_COUNT_SORT_OPTION ) ) {
 760                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for sorting of domain counts: -"
 761                         + surfacing.DOMAIN_COUNT_SORT_OPTION + "=<" + surfacing.DOMAIN_COUNT_SORT_ALPHA + "|"
 762                         + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
 763                         + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
 764                         + surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
 765             }
 766             final String sort = cla.getOptionValue( surfacing.DOMAIN_COUNT_SORT_OPTION ).toLowerCase();
 767             if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_ALPHA ) ) {
 768                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
 769             }
 770             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT ) ) {
 771                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_COUNT;
 772             }
 773             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT ) ) {
 774                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_PROTEINS_COUNT;
 775             }
 776             else if ( sort.equals( surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT ) ) {
 777                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.COMBINATIONS_COUNT;
 778             }
 779             else {
 780                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + sort
 781                                          + "\" for sorting of domain counts: \"-" + surfacing.DOMAIN_COUNT_SORT_OPTION + "=<"
 782                                          + surfacing.DOMAIN_COUNT_SORT_ALPHA + "|" + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
 783                                          + surfacing.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
 784                                          + surfacing.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
 785             }
 786         }
 787         final String[][] input_file_properties = SurfacingUtil.processInputGenomesFile( input_genomes_file );
 788         final int number_of_genomes = input_file_properties.length;
 789         if ( number_of_genomes < 2 ) {
 790             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot analyze less than two files" );
 791         }
 792         if ( ( number_of_genomes < 3 ) && perform_pwc ) {
 793             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use : -"
 794                     + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
 795                     + "=<suffix> to turn on pairwise analyses with less than three input files" );
 796         }
 797         SurfacingUtil.checkWriteabilityForPairwiseComparisons( domain_similarity_print_option,
 798                                                                input_file_properties,
 799                                                                automated_pairwise_comparison_suffix,
 800                                                                out_dir );
 801         for( int i = 0; i < number_of_genomes; i++ ) {
 802             File dcc_outfile = new File( input_file_properties[ i ][ 1 ]
 803                     + surfacing.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
 804             if ( out_dir != null ) {
 805                 dcc_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + dcc_outfile );
 806             }
 807             SurfacingUtil.checkForOutputFileWriteability( dcc_outfile );
 808         }
 809         File pfam_to_go_file = new File( "pfam2go.txt" );
 810         if ( cla.isOptionSet( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) ) {
 811             if ( !cla.isOptionValueSet( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) ) {
 812                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for Pfam to GO mapping file: -"
 813                         + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>" );
 814             }
 815             pfam_to_go_file = new File( cla.getOptionValue( surfacing.PFAM_TO_GO_FILE_USE_OPTION ) );
 816         }
 817         final String error1 = ForesterUtil.isReadableFile( pfam_to_go_file );
 818         if ( !ForesterUtil.isEmpty( error1 ) ) {
 819             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read Pfam to GO mapping file: " + error1 );
 820         }
 821         Map<String, List<GoId>> domain_id_to_go_ids_map = null;
 822         int domain_id_to_go_ids_count = 0;
 823         try {
 824             final PfamToGoParser parser = new PfamToGoParser( pfam_to_go_file );
 825             final List<PfamToGoMapping> pfam_to_go_mappings = parser.parse();
 826             domain_id_to_go_ids_map = SurfacingUtil.createDomainIdToGoIdMap( pfam_to_go_mappings );
 827             if ( parser.getMappingCount() < domain_id_to_go_ids_map.size() ) {
 828                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME,
 829                         "parser.getMappingCount() < domain_id_to_go_ids_map.size()" );
 830             }
 831             domain_id_to_go_ids_count = parser.getMappingCount();
 832         }
 833         catch ( final IOException e ) {
 834             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from Pfam to GO mapping file: " + e );
 835         }
 836         File go_obo_file = new File( "go.obo" );
 837         if ( cla.isOptionSet( surfacing.GO_OBO_FILE_USE_OPTION ) ) {
 838             if ( !cla.isOptionValueSet( surfacing.GO_OBO_FILE_USE_OPTION ) ) {
 839                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for GO OBO file: -"
 840                         + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>" );
 841             }
 842             go_obo_file = new File( cla.getOptionValue( surfacing.GO_OBO_FILE_USE_OPTION ) );
 843         }
 844         final String error2 = ForesterUtil.isReadableFile( go_obo_file );
 845         if ( !ForesterUtil.isEmpty( error2 ) ) {
 846             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read GO OBO file: " + error2 );
 847         }
 848         List<GoTerm> go_terms = null;
 849         try {
 850             final OBOparser parser = new OBOparser( go_obo_file, OBOparser.ReturnType.BASIC_GO_TERM );
 851             go_terms = parser.parse();
 852             if ( parser.getGoTermCount() != go_terms.size() ) {
 853                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "parser.getGoTermCount() != go_terms.size()" );
 854             }
 855         }
 856         catch ( final IOException e ) {
 857             ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read from GO OBO file: " + e );
 858         }
 859         Map<GoId, GoTerm> go_id_to_term_map = null;
 860         if ( ( ( domain_id_to_go_ids_map != null ) && ( domain_id_to_go_ids_map.size() > 0 ) )
 861                 && ( ( go_terms != null ) && ( go_terms.size() > 0 ) ) ) {
 862             go_id_to_term_map = GoUtils.createGoIdToGoTermMap( go_terms );
 863         }
 864         GoNameSpace go_namespace_limit = null;
 865         if ( cla.isOptionSet( surfacing.GO_NAMESPACE_LIMIT_OPTION ) ) {
 866             if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
 867                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot use GO namespace limit (-"
 868                         + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<namespace>) without Pfam to GO mapping file ("
 869                         + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>) and GO OBO file (-"
 870                         + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>)" );
 871             }
 872             if ( !cla.isOptionValueSet( surfacing.GO_NAMESPACE_LIMIT_OPTION ) ) {
 873                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for GO namespace limit: \"-"
 874                         + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<"
 875                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
 876                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
 877                         + surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
 878             }
 879             final String go_namespace_limit_str = cla.getOptionValue( surfacing.GO_NAMESPACE_LIMIT_OPTION )
 880                     .toLowerCase();
 881             if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION ) ) {
 882                 go_namespace_limit = GoNameSpace.createMolecularFunction();
 883             }
 884             else if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS ) ) {
 885                 go_namespace_limit = GoNameSpace.createBiologicalProcess();
 886             }
 887             else if ( go_namespace_limit_str.equals( surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT ) ) {
 888                 go_namespace_limit = GoNameSpace.createCellularComponent();
 889             }
 890             else {
 891                 ForesterUtil.fatalError( surfacing.PRG_NAME, "unknown value \"" + go_namespace_limit_str
 892                                          + "\" for GO namespace limit: \"-" + surfacing.GO_NAMESPACE_LIMIT_OPTION + "=<"
 893                                          + surfacing.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
 894                                          + surfacing.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
 895                                          + surfacing.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
 896             }
 897         }
 898         if ( ( domain_similarity_sort_field == DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE )
 899                 && ( number_of_genomes > 2 ) ) {
 900             domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 901         }
 902         File[] intree_files = null;
 903         Phylogeny[] intrees = null;
 904         if ( cla.isOptionSet( surfacing.INPUT_SPECIES_TREE_OPTION ) ) {
 905             if ( number_of_genomes < 3 ) {
 906                 ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot infer gains and losses on input species trees (-"
 907                         + surfacing.INPUT_SPECIES_TREE_OPTION + " without pairwise analyses ("
 908                         + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
 909                         + "=<suffix for pairwise comparison output files>)" );
 910             }
 911             if ( !cla.isOptionValueSet( surfacing.INPUT_SPECIES_TREE_OPTION ) ) {
 912                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for input tree: -"
 913                         + surfacing.INPUT_SPECIES_TREE_OPTION + "=<tree file in phyloXML format>" );
 914             }
 915             final String intrees_str = cla.getOptionValue( surfacing.INPUT_SPECIES_TREE_OPTION );
 916             if ( intrees_str.indexOf( "#" ) > 0 ) {
 917                 final String[] intrees_strs = intrees_str.split( "#" );
 918                 intree_files = new File[ intrees_strs.length ];
 919                 int i = 0;
 920                 for( final String s : intrees_strs ) {
 921                     intree_files[ i++ ] = new File( s.trim() );
 922                 }
 923             }
 924             else {
 925                 intree_files = new File[ 1 ];
 926                 intree_files[ 0 ] = new File( intrees_str );
 927             }
 928             intrees = SurfacingUtil.obtainAndPreProcessIntrees( intree_files, number_of_genomes, input_file_properties );
 929         }
 930         final Phylogeny intree_0_orig = SurfacingUtil.obtainFirstIntree( intree_files[ 0 ] );
 931         long random_number_seed_for_fitch_parsimony = 0l;
 932         boolean radomize_fitch_parsimony = false;
 933         if ( cla.isOptionSet( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
 934             if ( !cla.isOptionValueSet( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
 935                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for random number seed: -"
 936                         + surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION + "=<seed>" );
 937             }
 938             try {
 939                 random_number_seed_for_fitch_parsimony = cla
 940                         .getOptionValueAsLong( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
 941             }
 942             catch ( final IOException e ) {
 943                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
 944             }
 945             radomize_fitch_parsimony = true;
 946         }
 947         SortedSet<String> filter = null;
 948         if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
 949                 || ( negative_domains_filter_file != null ) ) {
 950             filter = new TreeSet<String>();
 951             if ( positive_filter_file != null ) {
 952                 SurfacingUtil.processFilter( positive_filter_file, filter );
 953             }
 954             else if ( negative_filter_file != null ) {
 955                 SurfacingUtil.processFilter( negative_filter_file, filter );
 956             }
 957             else if ( negative_domains_filter_file != null ) {
 958                 SurfacingUtil.processFilter( negative_domains_filter_file, filter );
 959             }
 960         }
 961         Map<String, Set<String>>[] domain_id_to_secondary_features_maps = null;
 962         File[] secondary_features_map_files = null;
 963         final File domain_lengths_analysis_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
 964                                                                + DOMAIN_LENGTHS_ANALYSIS_SUFFIX );
 965         if ( domain_length_analysis ) {
 966             SurfacingUtil.checkForOutputFileWriteability( domain_lengths_analysis_outfile );
 967         }
 968         if ( cla.isOptionSet( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
 969             if ( !cla.isOptionValueSet( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
 970                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no value for secondary features map file: -"
 971                         + surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE + "=<file>" );
 972             }
 973             final String[] secondary_features_map_files_strs = cla
 974                     .getOptionValue( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ).split( "#" );
 975             secondary_features_map_files = new File[ secondary_features_map_files_strs.length ];
 976             domain_id_to_secondary_features_maps = new Map[ secondary_features_map_files_strs.length ];
 977             int i = 0;
 978             for( final String secondary_features_map_files_str : secondary_features_map_files_strs ) {
 979                 secondary_features_map_files[ i ] = new File( secondary_features_map_files_str );
 980                 final String error = ForesterUtil.isReadableFile( secondary_features_map_files[ i ] );
 981                 if ( !ForesterUtil.isEmpty( error ) ) {
 982                     ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot read secondary features map file: " + error );
 983                 }
 984                 try {
 985                     domain_id_to_secondary_features_maps[ i ] = SurfacingUtil
 986                             .createDomainIdToSecondaryFeaturesMap( secondary_features_map_files[ i ] );
 987                 }
 988                 catch ( final IOException e ) {
 989                     ForesterUtil.fatalError( surfacing.PRG_NAME,
 990                                              "cannot read secondary features map file: " + e.getMessage() );
 991                 }
 992                 catch ( final Exception e ) {
 993                     ForesterUtil.fatalError( surfacing.PRG_NAME, "problem with contents of features map file ["
 994                             + secondary_features_map_files[ i ] + "]: " + e.getMessage() );
 995                 }
 996                 i++;
 997             }
 998         }
 999         if ( out_dir == null ) {
1000             ForesterUtil.fatalError( surfacing.PRG_NAME, "no output directory indicated (-"
1001                     + surfacing.OUTPUT_DIR_OPTION + "=<dir>)" );
1002         }
1003         if ( output_file == null ) {
1004             ForesterUtil.fatalError( surfacing.PRG_NAME, "no name for (main) output file indicated (-"
1005                     + surfacing.OUTPUT_FILE_OPTION + "=<file>)" );
1006         }
1007         if ( ( domain_id_to_go_ids_map == null ) || domain_id_to_go_ids_map.isEmpty() ) {
1008             ForesterUtil.fatalError( surfacing.PRG_NAME,
1009                                      "no (acceptable) Pfam to GO id mapping file provided ('pfam2go file') (-"
1010                                              + surfacing.PFAM_TO_GO_FILE_USE_OPTION + "=<file>)" );
1011         }
1012         if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
1013             ForesterUtil.fatalError( surfacing.PRG_NAME,
1014                                      "no (acceptable) go id to term mapping file provided ('GO OBO file') (-"
1015                                              + surfacing.GO_OBO_FILE_USE_OPTION + "=<file>)" );
1016         }
1017         System.out.println( "Output directory            : " + out_dir );
1018         System.out.println( "Input genomes from          : " + input_genomes_file );
1019         html_desc.append( "<tr><td>Input genomes from:</td><td>" + input_genomes_file + "</td></tr>" + nl );
1020         if ( positive_filter_file != null ) {
1021             final int filter_size = filter.size();
1022             System.out.println( "Positive protein filter     : " + positive_filter_file + " [" + filter_size
1023                                 + " domain ids]" );
1024             html_desc.append( "<tr><td>Positive protein filter:</td><td>" + positive_filter_file + " [" + filter_size
1025                               + " domain ids]</td></tr>" + nl );
1026         }
1027         if ( negative_filter_file != null ) {
1028             final int filter_size = filter.size();
1029             System.out.println( "Negative protein filter     : " + negative_filter_file + " [" + filter_size
1030                                 + " domain ids]" );
1031             html_desc.append( "<tr><td>Negative protein filter:</td><td>" + negative_filter_file + " [" + filter_size
1032                               + " domain ids]</td></tr>" + nl );
1033         }
1034         if ( negative_domains_filter_file != null ) {
1035             final int filter_size = filter.size();
1036             System.out.println( "Negative domain filter      : " + negative_domains_filter_file + " [" + filter_size
1037                                 + " domain ids]" );
1038             html_desc.append( "<tr><td>Negative domain filter:</td><td>" + negative_domains_filter_file + " ["
1039                     + filter_size + " domain ids]</td></tr>" + nl );
1040         }
1041         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
1042             String plus0 = "";
1043             for( final String s : plus_minus_analysis_high_copy_base_species ) {
1044                 plus0 += "+" + s + " ";
1045             }
1046             String plus1 = "";
1047             for( final String s : plus_minus_analysis_high_copy_target_species ) {
1048                 plus1 += "*" + s + " ";
1049             }
1050             String minus = "";
1051             for( final String s : plus_minus_analysis_high_low_copy_species ) {
1052                 minus += "-" + s + " ";
1053             }
1054             System.out.println( "Plus-minus analysis         : " + plus1 + "&& " + plus0 + "&& " + minus );
1055             html_desc.append( "<tr><td>Plus-minus analysis:</td><td>" + plus1 + "&& " + plus0 + "&& " + minus
1056                               + "</td></tr>" + nl );
1057         }
1058         if ( cutoff_scores_file != null ) {
1059             System.out.println( "Cutoff scores file          : " + cutoff_scores_file );
1060             html_desc.append( "<tr><td>Cutoff scores file:</td><td>" + cutoff_scores_file + "</td></tr>" + nl );
1061         }
1062         if ( ie_value_max >= 0.0 ) {
1063             System.out.println( "iE-value maximum (incl)     : " + ie_value_max );
1064             html_desc.append( "<tr><td>iE-value maximum (inclusive):</td><td>" + ie_value_max + "</td></tr>" + nl );
1065         }
1066         if ( fs_e_value_max >= 0.0 ) {
1067             System.out.println( "FS E-value maximum (incl)   : " + fs_e_value_max );
1068             html_desc.append( "<tr><td>FS E-value maximum (inclusive):</td><td>" + fs_e_value_max + "</td></tr>" + nl );
1069         }
1070         if ( output_protein_lists_for_all_domains ) {
1071             System.out.println( "Domain E-value max          : " + output_list_of_all_proteins_per_domain_e_value_max );
1072             html_desc.append( "<tr><td>Protein lists: E-value maximum per domain (inclusive):</td><td>"
1073                     + output_list_of_all_proteins_per_domain_e_value_max + "</td></tr>" + nl );
1074         }
1075         System.out.println( "Ignore DUFs                 : " + ignore_dufs );
1076         if ( ignore_virus_like_ids ) {
1077             System.out.println( "Ignore virus like ids       : " + ignore_virus_like_ids );
1078             html_desc.append( "<tr><td>Ignore virus, phage, transposition related ids:</td><td>"
1079                     + ignore_virus_like_ids + "</td></tr>" + nl );
1080         }
1081         html_desc.append( "<tr><td>Ignore DUFs:</td><td>" + ignore_dufs + "</td></tr>" + nl );
1082         if ( max_allowed_overlap != surfacing.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1083             System.out.println( "Max allowed domain overlap  : " + max_allowed_overlap );
1084             html_desc.append( "<tr><td>Max allowed domain overlap:</td><td>" + max_allowed_overlap + "</td></tr>" + nl );
1085         }
1086         if ( no_engulfing_overlaps ) {
1087             System.out.println( "Ignore engulfed domains     : " + no_engulfing_overlaps );
1088             html_desc.append( "<tr><td>Ignore (lower confidence) engulfed domains:</td><td>" + no_engulfing_overlaps
1089                               + "</td></tr>" + nl );
1090         }
1091         System.out.println( "Ignore singlet domains      : " + ignore_domains_without_combs_in_all_spec );
1092         html_desc
1093         .append( "<tr><td>Ignore singlet domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1094                 + ignore_domains_without_combs_in_all_spec + "</td></tr>" + nl );
1095         System.out.println( "Ignore species specific doms: " + ignore_species_specific_domains );
1096         html_desc
1097         .append( "<tr><td>Ignore species specific domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1098                 + ignore_species_specific_domains + "</td></tr>" + nl );
1099         System.out.println( "Ignore combination with self: " + ignore_combination_with_same );
1100         html_desc.append( "<tr><td>Ignore combination with self for domain combination similarity analyses:</td><td>"
1101                 + ignore_combination_with_same + "</td></tr>" + nl );
1102         System.out.println( "Consider directedness       : "
1103                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) );
1104         html_desc.append( "<tr><td>Consider directedness of binary domain combinations:</td><td>"
1105                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) + "</td></tr>" + nl );
1106         if ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) {
1107             System.out.println( "Consider adjacency          : "
1108                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) );
1109             html_desc.append( "<tr><td>Consider djacency of binary domain combinations:</td><td>"
1110                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) + "</td></tr>"
1111                     + nl );
1112         }
1113         System.out.println( "Fitch parsimony of DCs      : " + perform_dc_fich );
1114         html_desc.append( "<tr><td>Fitch parsimony of DCs:</td><td>" + perform_dc_fich + "</td></tr>" + nl );
1115         if ( perform_dc_fich ) {
1116             System.out.println( "Use last in Fitch parsimony : " + use_last_in_fitch_parsimony );
1117             html_desc.append( "<tr><td>Use last in Fitch parsimony:</td><td>" + use_last_in_fitch_parsimony
1118                               + "</td></tr>" + nl );
1119         }
1120         System.out.println( "Write to Nexus files        : " + write_to_nexus );
1121         html_desc.append( "<tr><td>Write to Nexus files:</td><td>" + write_to_nexus + "</td></tr>" + nl );
1122         if ( perform_dc_fich ) {
1123             System.out.println( "DC regain prot stats        : " + perform_dc_regain_proteins_stats );
1124             html_desc.append( "<tr><td>DC regain prot stats:</td><td>" + perform_dc_regain_proteins_stats
1125                               + "</td></tr>" + nl );
1126         }
1127         System.out.println( "DA analysis                 : " + da_analysis );
1128         html_desc.append( "<tr><td>DA analysis :</td><td>" + da_analysis + "</td></tr>" + nl );
1129         System.out.print( "Domain counts sort order    : " );
1130         html_desc.append( "<tr><td>Domain counts sort order:</td><td>" );
1131         switch ( dc_sort_order ) {
1132             case ALPHABETICAL_KEY_ID:
1133                 System.out.println( "alphabetical" );
1134                 html_desc.append( "alphabetical" + "</td></tr>" + nl );
1135                 break;
1136             case KEY_DOMAIN_COUNT:
1137                 System.out.println( "domain count" );
1138                 html_desc.append( "domain count" + "</td></tr>" + nl );
1139                 break;
1140             case KEY_DOMAIN_PROTEINS_COUNT:
1141                 System.out.println( "domain proteins count" );
1142                 html_desc.append( "domain proteins count" + "</td></tr>" + nl );
1143                 break;
1144             case COMBINATIONS_COUNT:
1145                 System.out.println( "domain combinations count" );
1146                 html_desc.append( "domain combinations count" + "</td></tr>" + nl );
1147                 break;
1148             default:
1149                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for dc sort order" );
1150         }
1151         if ( domain_id_to_go_ids_map != null ) {
1152             System.out.println( "Pfam to GO mappings from    : " + pfam_to_go_file + " [" + domain_id_to_go_ids_count
1153                                 + " mappings]" );
1154             html_desc.append( "<tr><td>Pfam to GO mappings from:</td><td>" + pfam_to_go_file + " ["
1155                     + domain_id_to_go_ids_count + " mappings]" + "</td></tr>" + nl );
1156         }
1157         if ( go_terms != null ) {
1158             System.out.println( "GO terms from               : " + go_obo_file + " [" + go_terms.size() + " terms]" );
1159             html_desc.append( "<tr><td>GO terms from:</td><td>" + go_obo_file + " [" + go_terms.size() + " terms]"
1160                     + "</td></tr>" + nl );
1161         }
1162         if ( go_namespace_limit != null ) {
1163             System.out.println( "Limit GO terms to           : " + go_namespace_limit.toString() );
1164             html_desc.append( "<tr><td>Limit GO terms to</td><td>" + go_namespace_limit + "</td></tr>" + nl );
1165         }
1166         if ( perform_pwc ) {
1167             System.out.println( "Suffix for PWC files        : " + automated_pairwise_comparison_suffix );
1168             html_desc.append( "<tr><td>Suffix for PWC files</td><td>" + automated_pairwise_comparison_suffix
1169                               + "</td></tr>" + nl );
1170         }
1171         if ( out_dir != null ) {
1172             System.out.println( "Output directory            : " + out_dir );
1173         }
1174         if ( query_domain_ids != null ) {
1175             System.out.println( "Query domains (ordered)     : " + query_domain_ids );
1176             html_desc.append( "<tr><td></td><td>" + query_domain_ids + "</td></tr>" + nl );
1177         }
1178         System.out.println( "Write similarities to       : " + output_file );
1179         System.out.print( "  Scoring method            : " );
1180         html_desc.append( "<tr><td>Scoring method:</td><td>" );
1181         switch ( scoring ) {
1182             case COMBINATIONS:
1183                 System.out.println( "domain combinations based" );
1184                 html_desc.append( "domain combinations based" + "</td></tr>" + nl );
1185                 break;
1186             case DOMAINS:
1187                 System.out.println( "domain counts based" );
1188                 html_desc.append( "domain counts based" + "</td></tr>" + nl );
1189                 break;
1190             case PROTEINS:
1191                 System.out.println( "domain proteins counts based" );
1192                 html_desc.append( "domain proteins counts based" + "</td></tr>" + nl );
1193                 break;
1194             default:
1195                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for scoring" );
1196         }
1197         System.out.print( "  Sort by                   : " );
1198         html_desc.append( "<tr><td>Sort by:</td><td>" );
1199         switch ( domain_similarity_sort_field ) {
1200             case MIN:
1201                 System.out.print( "score minimum" );
1202                 html_desc.append( "score minimum" );
1203                 break;
1204             case MAX:
1205                 System.out.print( "score maximum" );
1206                 html_desc.append( "score maximum" );
1207                 break;
1208             case MEAN:
1209                 System.out.print( "score mean" );
1210                 html_desc.append( "score mean" );
1211                 break;
1212             case SD:
1213                 System.out.print( "score standard deviation" );
1214                 html_desc.append( "score standard deviation" );
1215                 break;
1216             case SPECIES_COUNT:
1217                 System.out.print( "species number" );
1218                 html_desc.append( "species number" );
1219                 break;
1220             case DOMAIN_ID:
1221                 System.out.print( "alphabetical domain identifier" );
1222                 html_desc.append( "alphabetical domain identifier" );
1223                 break;
1224             case MAX_DIFFERENCE:
1225                 System.out.print( "(maximal) difference" );
1226                 html_desc.append( "(maximal) difference" );
1227                 break;
1228             case ABS_MAX_COUNTS_DIFFERENCE:
1229                 System.out.print( "absolute (maximal) counts difference" );
1230                 html_desc.append( "absolute (maximal) counts difference" );
1231                 break;
1232             case MAX_COUNTS_DIFFERENCE:
1233                 System.out.print( "(maximal) counts difference" );
1234                 html_desc.append( "(maximal) counts  difference" );
1235                 break;
1236             default:
1237                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for similarities" );
1238         }
1239         if ( sort_by_species_count_first ) {
1240             System.out.println( " (sort by species count first)" );
1241             html_desc.append( " (sort by species count first)" );
1242         }
1243         else {
1244             System.out.println();
1245         }
1246         html_desc.append( "</td></tr>" + nl );
1247         System.out.print( "  Detailedness              : " );
1248         switch ( detailedness ) {
1249             case BASIC:
1250                 System.out.println( "basic" );
1251                 break;
1252             case LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES:
1253                 System.out.println( "list combining domains for each species" );
1254                 break;
1255             case PUNCTILIOUS:
1256                 System.out.println( "punctilious" );
1257                 break;
1258             default:
1259                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for detailedness" );
1260         }
1261         System.out.print( "  Print option              : " );
1262         switch ( domain_similarity_print_option ) {
1263             case HTML:
1264                 System.out.println( "HTML" );
1265                 break;
1266             case SIMPLE_TAB_DELIMITED:
1267                 System.out.println( "simple tab delimited" );
1268                 break;
1269             default:
1270                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for print option" );
1271         }
1272         System.out.print( "  Species matrix            : " + species_matrix );
1273         System.out.println();
1274         final File dc_data_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file + DATA_FILE_SUFFIX );
1275         System.out.println( "Domain comb data output     : " + dc_data_file );
1276         html_desc.append( "<tr><td>Domain combination data output:</td><td> " + dc_data_file + " </td></tr>" );
1277         System.out.println();
1278         if ( perform_pwc ) {
1279             System.out.println( "Pairwise comparisons: " );
1280             html_desc.append( "<tr><td>Pairwise comparisons:</td><td></td></tr>" );
1281             System.out.print( "  Sort by                   : " );
1282             html_desc.append( "<tr><td>Sort by:</td><td>" );
1283             switch ( domain_similarity_sort_field_for_automated_pwc ) {
1284                 case MEAN:
1285                     System.out.print( "score mean" );
1286                     html_desc.append( "score mean" );
1287                     break;
1288                 case DOMAIN_ID:
1289                     System.out.print( "alphabetical domain identifier" );
1290                     html_desc.append( "alphabetical domain identifier" );
1291                     break;
1292                 case MAX_DIFFERENCE:
1293                     System.out.print( "difference" );
1294                     html_desc.append( "difference" );
1295                     break;
1296                 case ABS_MAX_COUNTS_DIFFERENCE:
1297                     System.out.print( "absolute counts difference" );
1298                     html_desc.append( "absolute counts difference" );
1299                     break;
1300                 case MAX_COUNTS_DIFFERENCE:
1301                     System.out.print( "counts difference" );
1302                     html_desc.append( "counts difference" );
1303                     break;
1304                 default:
1305                     ForesterUtil
1306                     .unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for similarities" );
1307             }
1308             System.out.println();
1309             html_desc.append( "</td></tr>" + nl );
1310             if ( ( intrees != null ) && ( intrees.length > 0 ) ) {
1311                 for( final File intree_file : intree_files ) {
1312                     html_desc.append( "<tr><td>Intree for gain/loss parsimony analysis:</td><td>" + intree_file
1313                                       + "</td></tr>" + nl );
1314                     System.out.println( "  Intree for gain/loss pars.: " + intree_file );
1315                 }
1316             }
1317             if ( radomize_fitch_parsimony ) {
1318                 html_desc.append( "<tr><td>    Random number seed for Fitch parsimony analysis:</td><td>"
1319                         + random_number_seed_for_fitch_parsimony + "</td></tr>" + nl );
1320                 System.out.println( "    Random number seed      : " + random_number_seed_for_fitch_parsimony );
1321             }
1322             if ( ( domain_id_to_secondary_features_maps != null ) && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
1323                 for( int i = 0; i < secondary_features_map_files.length; i++ ) {
1324                     html_desc.append( "<tr><td>Secondary features map file:</td><td>"
1325                             + secondary_features_map_files[ i ] + "</td></tr>" + nl );
1326                     System.out.println( "Secondary features map file : " + secondary_features_map_files[ i ]
1327                             + " [mappings for " + domain_id_to_secondary_features_maps[ i ].size() + " domain ids]" );
1328                     if ( VERBOSE ) {
1329                         System.out.println();
1330                         System.out.println( "Domain ids to secondary features map:" );
1331                         for( final String domain_id : domain_id_to_secondary_features_maps[ i ].keySet() ) {
1332                             System.out.print( domain_id );
1333                             System.out.print( " => " );
1334                             for( final String sec : domain_id_to_secondary_features_maps[ i ].get( domain_id ) ) {
1335                                 System.out.print( sec );
1336                                 System.out.print( " " );
1337                             }
1338                             System.out.println();
1339                         }
1340                     }
1341                 }
1342             }
1343         } // if ( perform_pwc ) {
1344         System.out.println();
1345         html_desc.append( "<tr><td>Command line:</td><td>" + nl + nl + cla.getCommandLineArgsAsString() + nl + nl
1346                           + "</td></tr>" + nl );
1347         System.out.println( "Command line                : " + cla.getCommandLineArgsAsString() );
1348         BufferedWriter[] query_domains_writer_ary = null;
1349         List<String>[] query_domain_ids_array = null;
1350         if ( query_domain_ids != null ) {
1351             final String[] query_domain_ids_str_array = query_domain_ids.split( "#" );
1352             query_domain_ids_array = new ArrayList[ query_domain_ids_str_array.length ];
1353             query_domains_writer_ary = new BufferedWriter[ query_domain_ids_str_array.length ];
1354             for( int i = 0; i < query_domain_ids_str_array.length; i++ ) {
1355                 String query_domain_ids_str = query_domain_ids_str_array[ i ];
1356                 final String[] query_domain_ids_str_ary = query_domain_ids_str.split( "~" );
1357                 final List<String> query = new ArrayList<String>();
1358                 for( final String element : query_domain_ids_str_ary ) {
1359                     query.add( element );
1360                 }
1361                 query_domain_ids_array[ i ] = query;
1362                 query_domain_ids_str = query_domain_ids_str.replace( '~', '_' );
1363                 String protein_names_writer_str = query_domain_ids_str + surfacing.SEQ_EXTRACT_SUFFIX;
1364                 if ( out_dir != null ) {
1365                     protein_names_writer_str = out_dir + ForesterUtil.FILE_SEPARATOR + protein_names_writer_str;
1366                 }
1367                 try {
1368                     query_domains_writer_ary[ i ] = new BufferedWriter( new FileWriter( protein_names_writer_str ) );
1369                 }
1370                 catch ( final IOException e ) {
1371                     ForesterUtil.fatalError( surfacing.PRG_NAME, "Could not open [" + protein_names_writer_str + "]: "
1372                             + e.getLocalizedMessage() );
1373                 }
1374             }
1375         }
1376         SortedMap<Species, List<Protein>> protein_lists_per_species = null; //This will only be created if neede.
1377         boolean need_protein_lists_per_species = false;
1378         if ( ( plus_minus_analysis_high_copy_base_species.size() > 0 ) || output_protein_lists_for_all_domains ) {
1379             need_protein_lists_per_species = true;
1380         }
1381         if ( need_protein_lists_per_species ) {
1382             protein_lists_per_species = new TreeMap<Species, List<Protein>>();
1383         }
1384         List<GenomeWideCombinableDomains> gwcd_list = new ArrayList<GenomeWideCombinableDomains>( number_of_genomes );
1385         final SortedSet<String> all_domains_encountered = new TreeSet<String>();
1386         final SortedSet<BinaryDomainCombination> all_bin_domain_combinations_encountered = new TreeSet<BinaryDomainCombination>();
1387         List<BinaryDomainCombination> all_bin_domain_combinations_gained_fitch = null;
1388         List<BinaryDomainCombination> all_bin_domain_combinations_lost_fitch = null;
1389         if ( ( intrees != null ) && ( intrees.length == 1 ) ) {
1390             all_bin_domain_combinations_gained_fitch = new ArrayList<BinaryDomainCombination>();
1391             all_bin_domain_combinations_lost_fitch = new ArrayList<BinaryDomainCombination>();
1392         }
1393         final File per_genome_domain_promiscuity_statistics_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR
1394                                                                              + output_file + D_PROMISCUITY_FILE_SUFFIX );
1395         BufferedWriter per_genome_domain_promiscuity_statistics_writer = null;
1396         try {
1397             per_genome_domain_promiscuity_statistics_writer = new BufferedWriter( new FileWriter( per_genome_domain_promiscuity_statistics_file ) );
1398             per_genome_domain_promiscuity_statistics_writer.write( "Species:\t" );
1399             per_genome_domain_promiscuity_statistics_writer.write( "Mean:\t" );
1400             per_genome_domain_promiscuity_statistics_writer.write( "SD:\t" );
1401             per_genome_domain_promiscuity_statistics_writer.write( "Median:\t" );
1402             per_genome_domain_promiscuity_statistics_writer.write( "Min:\t" );
1403             per_genome_domain_promiscuity_statistics_writer.write( "Max:\t" );
1404             per_genome_domain_promiscuity_statistics_writer.write( "N:\t" );
1405             per_genome_domain_promiscuity_statistics_writer.write( "Max Promiscuous Domains:"
1406                     + ForesterUtil.LINE_SEPARATOR );
1407         }
1408         catch ( final IOException e2 ) {
1409             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1410         }
1411         final File log_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file + LOG_FILE_SUFFIX );
1412         BufferedWriter log_writer = null;
1413         try {
1414             log_writer = new BufferedWriter( new FileWriter( log_file ) );
1415         }
1416         catch ( final IOException e2 ) {
1417             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1418         }
1419         BufferedWriter dc_data_writer = null;
1420         try {
1421             dc_data_writer = new BufferedWriter( new FileWriter( dc_data_file ) );
1422             dc_data_writer.write( DATA_FILE_DESC );
1423             dc_data_writer.write( ForesterUtil.LINE_SEPARATOR );
1424         }
1425         catch ( final IOException e2 ) {
1426             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
1427         }
1428         DescriptiveStatistics protein_coverage_stats = new BasicDescriptiveStatistics();
1429         DescriptiveStatistics all_genomes_domains_per_potein_stats = new BasicDescriptiveStatistics();
1430         final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo = new TreeMap<Integer, Integer>();
1431         final SortedSet<String> domains_which_are_always_single = new TreeSet<String>();
1432         final SortedSet<String> domains_which_are_sometimes_single_sometimes_not = new TreeSet<String>();
1433         final SortedSet<String> domains_which_never_single = new TreeSet<String>();
1434         BufferedWriter domains_per_potein_stats_writer = null;
1435         try {
1436             domains_per_potein_stats_writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR
1437                                                                                   + output_file + "_domains_per_potein_stats.txt" ) );
1438             domains_per_potein_stats_writer.write( "Genome" );
1439             domains_per_potein_stats_writer.write( "\t" );
1440             domains_per_potein_stats_writer.write( "Mean" );
1441             domains_per_potein_stats_writer.write( "\t" );
1442             domains_per_potein_stats_writer.write( "SD" );
1443             domains_per_potein_stats_writer.write( "\t" );
1444             domains_per_potein_stats_writer.write( "Median" );
1445             domains_per_potein_stats_writer.write( "\t" );
1446             domains_per_potein_stats_writer.write( "N" );
1447             domains_per_potein_stats_writer.write( "\t" );
1448             domains_per_potein_stats_writer.write( "Min" );
1449             domains_per_potein_stats_writer.write( "\t" );
1450             domains_per_potein_stats_writer.write( "Max" );
1451             domains_per_potein_stats_writer.write( "\n" );
1452         }
1453         catch ( final IOException e3 ) {
1454             e3.printStackTrace();
1455         }
1456         Map<String, DescriptiveStatistics> protein_length_stats_by_dc = null;
1457         Map<String, DescriptiveStatistics> domain_number_stats_by_dc = null;
1458         final Map<String, DescriptiveStatistics> domain_length_stats_by_domain = new HashMap<String, DescriptiveStatistics>();
1459         if ( perform_dc_regain_proteins_stats ) {
1460             protein_length_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
1461             domain_number_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
1462         }
1463         DomainLengthsTable domain_lengths_table = null;
1464         if ( domain_length_analysis ) {
1465             domain_lengths_table = new DomainLengthsTable();
1466         }
1467         // Main loop:
1468         final SortedMap<String, Set<String>> distinct_domain_architecutures_per_genome = new TreeMap<String, Set<String>>();
1469         final SortedMap<String, Integer> distinct_domain_architecuture_counts = new TreeMap<String, Integer>();
1470         for( int i = 0; i < number_of_genomes; ++i ) {
1471             System.out.println();
1472             System.out.println( ( i + 1 ) + "/" + number_of_genomes );
1473             SurfacingUtil.log( ( i + 1 ) + "/" + number_of_genomes, log_writer );
1474             System.out.println( "Processing                                     : " + input_file_properties[ i ][ 1 ]
1475                     + " [" + input_file_properties[ i ][ 0 ] + "]" );
1476             SurfacingUtil.log( "Genome                                         : " + input_file_properties[ i ][ 1 ]
1477                     + " [" + input_file_properties[ i ][ 0 ] + "]", log_writer );
1478             HmmscanPerDomainTableParser parser = null;
1479             INDIVIDUAL_SCORE_CUTOFF ind_score_cutoff = INDIVIDUAL_SCORE_CUTOFF.NONE;
1480             if ( individual_score_cutoffs != null ) {
1481                 ind_score_cutoff = INDIVIDUAL_SCORE_CUTOFF_DEFAULT;
1482             }
1483             if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
1484                     || ( negative_domains_filter_file != null ) ) {
1485                 HmmscanPerDomainTableParser.FilterType filter_type = HmmscanPerDomainTableParser.FilterType.NONE;
1486                 if ( positive_filter_file != null ) {
1487                     filter_type = HmmscanPerDomainTableParser.FilterType.POSITIVE_PROTEIN;
1488                 }
1489                 else if ( negative_filter_file != null ) {
1490                     filter_type = HmmscanPerDomainTableParser.FilterType.NEGATIVE_PROTEIN;
1491                 }
1492                 else if ( negative_domains_filter_file != null ) {
1493                     filter_type = HmmscanPerDomainTableParser.FilterType.NEGATIVE_DOMAIN;
1494                 }
1495                 parser = new HmmscanPerDomainTableParser( new File( input_file_properties[ i ][ 0 ] ),
1496                                                           input_file_properties[ i ][ 1 ],
1497                                                           filter,
1498                                                           filter_type,
1499                                                           ind_score_cutoff,
1500                                                           true );
1501             }
1502             else {
1503                 parser = new HmmscanPerDomainTableParser( new File( input_file_properties[ i ][ 0 ] ),
1504                                                           input_file_properties[ i ][ 1 ],
1505                                                           ind_score_cutoff,
1506                                                           true );
1507             }
1508             if ( fs_e_value_max >= 0.0 ) {
1509                 parser.setFsEValueMaximum( fs_e_value_max );
1510             }
1511             if ( ie_value_max >= 0.0 ) {
1512                 parser.setIEValueMaximum( ie_value_max );
1513             }
1514             parser.setIgnoreDufs( ignore_dufs );
1515             parser.setIgnoreVirusLikeIds( ignore_virus_like_ids );
1516             parser.setIgnoreEngulfedDomains( no_engulfing_overlaps );
1517             if ( max_allowed_overlap != surfacing.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1518                 parser.setMaxAllowedOverlap( max_allowed_overlap );
1519             }
1520             parser.setReturnType( HmmscanPerDomainTableParser.ReturnType.UNORDERED_PROTEIN_DOMAIN_COLLECTION_PER_PROTEIN );
1521             if ( individual_score_cutoffs != null ) {
1522                 parser.setIndividualScoreCutoffs( individual_score_cutoffs );
1523             }
1524             List<Protein> protein_list = null;
1525             try {
1526                 protein_list = parser.parse();
1527             }
1528             catch ( final IOException e ) {
1529                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1530             }
1531             catch ( final Exception e ) {
1532                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, e.getMessage(), e );
1533             }
1534             if ( VERBOSE ) {
1535                 System.out.println( "Domains ignored due to negative domain filter: " );
1536                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToNegativeDomainFilterCountsMap() );
1537                 System.out.println( "Domains ignored due to virus like id: " );
1538                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToVirusLikeIdCountsMap() );
1539             }
1540             final double coverage = ( double ) protein_list.size() / parser.getProteinsEncountered();
1541             protein_coverage_stats.addValue( coverage );
1542             int distinct_das = -1;
1543             if ( da_analysis ) {
1544                 final String genome = input_file_properties[ i ][ 0 ];
1545                 distinct_das = SurfacingUtil.storeDomainArchitectures( genome,
1546                                                                        distinct_domain_architecutures_per_genome,
1547                                                                        protein_list,
1548                                                                        distinct_domain_architecuture_counts );
1549             }
1550             System.out.println( "Number of proteins encountered                 : " + parser.getProteinsEncountered() );
1551             SurfacingUtil.log( "Number of proteins encountered                 : " + parser.getProteinsEncountered(),
1552                                log_writer );
1553             System.out.println( "Number of proteins stored                      : " + protein_list.size() );
1554             SurfacingUtil.log( "Number of proteins stored                      : " + protein_list.size(), log_writer );
1555             System.out.println( "Coverage                                       : "
1556                     + ForesterUtil.roundToInt( 100.0 * coverage ) + "%" );
1557             SurfacingUtil.log( "Coverage                                       : "
1558                     + ForesterUtil.roundToInt( 100.0 * coverage ) + "%",
1559                     log_writer );
1560             System.out.println( "Domains encountered                            : " + parser.getDomainsEncountered() );
1561             SurfacingUtil.log( "Domains encountered                            : " + parser.getDomainsEncountered(),
1562                                log_writer );
1563             System.out.println( "Domains stored                                 : " + parser.getDomainsStored() );
1564             SurfacingUtil.log( "Domains stored                                 : " + parser.getDomainsStored(),
1565                                log_writer );
1566             System.out.println( "Distinct domains stored                        : "
1567                     + parser.getDomainsStoredSet().size() );
1568             SurfacingUtil.log( "Distinct domains stored                        : "
1569                     + parser.getDomainsStoredSet().size(), log_writer );
1570             System.out.println( "Domains ignored due to individual score cutoffs: "
1571                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff() );
1572             SurfacingUtil.log( "Domains ignored due to individual score cutoffs: "
1573                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff(),
1574                     log_writer );
1575             System.out.println( "Domains ignored due to FS E-value              : "
1576                     + parser.getDomainsIgnoredDueToFsEval() );
1577             SurfacingUtil.log( "Domains ignored due to FS E-value              : "
1578                     + parser.getDomainsIgnoredDueToFsEval(),
1579                     log_writer );
1580             System.out.println( "Domains ignored due to iE-value                : "
1581                     + parser.getDomainsIgnoredDueToIEval() );
1582             SurfacingUtil.log( "Domains ignored due to iE-value                : "
1583                     + parser.getDomainsIgnoredDueToIEval(),
1584                     log_writer );
1585             System.out.println( "Domains ignored due to DUF designation         : "
1586                     + parser.getDomainsIgnoredDueToDuf() );
1587             SurfacingUtil
1588             .log( "Domains ignored due to DUF designation         : " + parser.getDomainsIgnoredDueToDuf(),
1589                   log_writer );
1590             if ( ignore_virus_like_ids ) {
1591                 System.out.println( "Domains ignored due virus like ids             : "
1592                         + parser.getDomainsIgnoredDueToVirusLikeIds() );
1593                 SurfacingUtil.log( "Domains ignored due virus like ids             : "
1594                         + parser.getDomainsIgnoredDueToVirusLikeIds(),
1595                         log_writer );
1596             }
1597             System.out.println( "Domains ignored due negative domain filter     : "
1598                     + parser.getDomainsIgnoredDueToNegativeDomainFilter() );
1599             SurfacingUtil.log( "Domains ignored due negative domain filter     : "
1600                     + parser.getDomainsIgnoredDueToNegativeDomainFilter(),
1601                     log_writer );
1602             System.out.println( "Domains ignored due to overlap                 : "
1603                     + parser.getDomainsIgnoredDueToOverlap() );
1604             SurfacingUtil.log( "Domains ignored due to overlap                 : "
1605                     + parser.getDomainsIgnoredDueToOverlap(),
1606                     log_writer );
1607             if ( negative_filter_file != null ) {
1608                 System.out.println( "Proteins ignored due to negative filter        : "
1609                         + parser.getProteinsIgnoredDueToFilter() );
1610                 SurfacingUtil.log( "Proteins ignored due to negative filter        : "
1611                         + parser.getProteinsIgnoredDueToFilter(),
1612                         log_writer );
1613             }
1614             if ( positive_filter_file != null ) {
1615                 System.out.println( "Proteins ignored due to positive filter        : "
1616                         + parser.getProteinsIgnoredDueToFilter() );
1617                 SurfacingUtil.log( "Proteins ignored due to positive filter        : "
1618                         + parser.getProteinsIgnoredDueToFilter(),
1619                         log_writer );
1620             }
1621             if ( da_analysis ) {
1622                 System.out.println( "Distinct domain architectures stored           : " + distinct_das );
1623                 SurfacingUtil.log( "Distinct domain architectures stored           : " + distinct_das, log_writer );
1624             }
1625             System.out.println( "Time for processing                            : " + parser.getTime() + "ms" );
1626             SurfacingUtil.log( "", log_writer );
1627             try {
1628                 int count = 0;
1629                 for( final Protein protein : protein_list ) {
1630                     dc_data_writer.write( SurfacingUtil.proteinToDomainCombinations( protein, count + "", "\t" )
1631                                           .toString() );
1632                     ++count;
1633                     for( final Domain d : protein.getProteinDomains() ) {
1634                         final String d_str = d.getDomainId().toString();
1635                         if ( !domain_length_stats_by_domain.containsKey( d_str ) ) {
1636                             domain_length_stats_by_domain.put( d_str, new BasicDescriptiveStatistics() );
1637                         }
1638                         domain_length_stats_by_domain.get( d_str ).addValue( d.getLength() );
1639                     }
1640                 }
1641             }
1642             catch ( final IOException e ) {
1643                 ForesterUtil.fatalError( surfacing.PRG_NAME, e.toString() );
1644             }
1645             SurfacingUtil.domainsPerProteinsStatistics( input_file_properties[ i ][ 1 ],
1646                                                         protein_list,
1647                                                         all_genomes_domains_per_potein_stats,
1648                                                         all_genomes_domains_per_potein_histo,
1649                                                         domains_which_are_always_single,
1650                                                         domains_which_are_sometimes_single_sometimes_not,
1651                                                         domains_which_never_single,
1652                                                         domains_per_potein_stats_writer );
1653             if ( domain_length_analysis ) {
1654                 domain_lengths_table.addLengths( protein_list );
1655             }
1656             if ( !da_analysis ) {
1657                 gwcd_list.add( BasicGenomeWideCombinableDomains
1658                                .createInstance( protein_list,
1659                                                 ignore_combination_with_same,
1660                                                 new BasicSpecies( input_file_properties[ i ][ 1 ] ),
1661                                                 domain_id_to_go_ids_map,
1662                                                 dc_type,
1663                                                 protein_length_stats_by_dc,
1664                                                 domain_number_stats_by_dc ) );
1665                 if ( gwcd_list.get( i ).getSize() > 0 ) {
1666                     if ( output_binary_domain_combinationsfor_counts ) {
1667                         SurfacingUtil
1668                         .writeDomainCombinationsCountsFile( input_file_properties,
1669                                                             out_dir,
1670                                                             per_genome_domain_promiscuity_statistics_writer,
1671                                                             gwcd_list.get( i ),
1672                                                             i,
1673                                                             dc_sort_order );
1674                     }
1675                     if ( output_binary_domain_combinationsfor_graph_analysis ) {
1676                         SurfacingUtil.writeBinaryDomainCombinationsFileForGraphAnalysis( input_file_properties,
1677                                                                                          out_dir,
1678                                                                                          gwcd_list.get( i ),
1679                                                                                          i,
1680                                                                                          dc_sort_order );
1681                     }
1682                     SurfacingUtil.addAllDomainIdsToSet( gwcd_list.get( i ), all_domains_encountered );
1683                     SurfacingUtil.addAllBinaryDomainCombinationToSet( gwcd_list.get( i ),
1684                                                                       all_bin_domain_combinations_encountered );
1685                 }
1686             }
1687             if ( query_domains_writer_ary != null ) {
1688                 for( int j = 0; j < query_domain_ids_array.length; j++ ) {
1689                     try {
1690                         SurfacingUtil.extractProteinNames( protein_list,
1691                                                            query_domain_ids_array[ j ],
1692                                                            query_domains_writer_ary[ j ],
1693                                                            "\t",
1694                                                            LIMIT_SPEC_FOR_PROT_EX );
1695                         query_domains_writer_ary[ j ].flush();
1696                     }
1697                     catch ( final IOException e ) {
1698                         e.printStackTrace();
1699                     }
1700                 }
1701             }
1702             if ( need_protein_lists_per_species ) {
1703                 protein_lists_per_species.put( new BasicSpecies( input_file_properties[ i ][ 1 ] ), protein_list );
1704             }
1705             try {
1706                 log_writer.flush();
1707             }
1708             catch ( final IOException e2 ) {
1709                 ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
1710             }
1711             System.gc();
1712         } // for( int i = 0; i < number_of_genomes; ++i ) {
1713         ForesterUtil.programMessage( PRG_NAME, "Wrote domain promiscuities to: "
1714                 + per_genome_domain_promiscuity_statistics_file );
1715         //
1716         if ( da_analysis ) {
1717             SurfacingUtil.performDomainArchitectureAnalysis( distinct_domain_architecutures_per_genome,
1718                                                              distinct_domain_architecuture_counts,
1719                                                              10,
1720                                                              new File( out_dir.toString() + "/" + output_file
1721                                                                        + "_DA_counts.txt" ),
1722                                                                        new File( out_dir.toString() + "/" + output_file
1723                                                                                  + "_unique_DAs.txt" ) );
1724             distinct_domain_architecutures_per_genome.clear();
1725             distinct_domain_architecuture_counts.clear();
1726             System.gc();
1727         }
1728         try {
1729             domains_per_potein_stats_writer.write( "ALL" );
1730             domains_per_potein_stats_writer.write( "\t" );
1731             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.arithmeticMean() + "" );
1732             domains_per_potein_stats_writer.write( "\t" );
1733             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.sampleStandardDeviation() + "" );
1734             domains_per_potein_stats_writer.write( "\t" );
1735             if ( all_genomes_domains_per_potein_stats.getN() <= 300 ) {
1736                 domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.median() + "" );
1737                 domains_per_potein_stats_writer.write( "\t" );
1738             }
1739             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getN() + "" );
1740             domains_per_potein_stats_writer.write( "\t" );
1741             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMin() + "" );
1742             domains_per_potein_stats_writer.write( "\t" );
1743             domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMax() + "" );
1744             domains_per_potein_stats_writer.write( "\n" );
1745             domains_per_potein_stats_writer.close();
1746             all_genomes_domains_per_potein_stats = null;
1747             SurfacingUtil.printOutPercentageOfMultidomainProteins( all_genomes_domains_per_potein_histo, log_writer );
1748             ForesterUtil.map2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1749                                              + "_all_genomes_domains_per_potein_histo.txt" ), all_genomes_domains_per_potein_histo, "\t", "\n" );
1750             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1751                                                     + "_domains_always_single_.txt" ), domains_which_are_always_single, "\n" );
1752             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1753                                                     + "_domains_single_or_combined.txt" ), domains_which_are_sometimes_single_sometimes_not, "\n" );
1754             ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1755                                                     + "_domains_always_combined.txt" ), domains_which_never_single, "\n" );
1756             ForesterUtil.programMessage( PRG_NAME,
1757                                          "Average of proteins with a least one domain assigned: "
1758                                                  + ( 100 * protein_coverage_stats.arithmeticMean() ) + "% (+/-"
1759                                                  + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)" );
1760             ForesterUtil.programMessage( PRG_NAME, "Range of proteins with a least one domain assigned: "
1761                     + ( 100 * protein_coverage_stats.getMin() ) + "%-" + ( 100 * protein_coverage_stats.getMax() )
1762                     + "%" );
1763             SurfacingUtil.log( "Average of prot with a least one dom assigned  : "
1764                     + ( 100 * protein_coverage_stats.arithmeticMean() ) + "% (+/-"
1765                     + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)",
1766                     log_writer );
1767             SurfacingUtil.log( "Range of prot with a least one dom assigned    : "
1768                     + ( 100 * protein_coverage_stats.getMin() ) + "%-"
1769                     + ( 100 * protein_coverage_stats.getMax() ) + "%",
1770                     log_writer );
1771             protein_coverage_stats = null;
1772         }
1773         catch ( final IOException e2 ) {
1774             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
1775         }
1776         if ( query_domains_writer_ary != null ) {
1777             for( int j = 0; j < query_domain_ids_array.length; j++ ) {
1778                 try {
1779                     query_domains_writer_ary[ j ].close();
1780                 }
1781                 catch ( final IOException e ) {
1782                     ForesterUtil.fatalError( surfacing.PRG_NAME, e.toString() );
1783                 }
1784             }
1785         }
1786         try {
1787             per_genome_domain_promiscuity_statistics_writer.close();
1788             dc_data_writer.close();
1789             log_writer.close();
1790         }
1791         catch ( final IOException e2 ) {
1792             ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
1793         }
1794         if ( domain_length_analysis ) {
1795             try {
1796                 SurfacingUtil.executeDomainLengthAnalysis( input_file_properties,
1797                                                            number_of_genomes,
1798                                                            domain_lengths_table,
1799                                                            domain_lengths_analysis_outfile );
1800             }
1801             catch ( final IOException e1 ) {
1802                 ForesterUtil.fatalError( surfacing.PRG_NAME, e1.toString() );
1803             }
1804             System.out.println();
1805             ForesterUtil.programMessage( PRG_NAME, "Wrote domain length data to: " + domain_lengths_analysis_outfile );
1806             System.out.println();
1807         }
1808         domain_lengths_table = null;
1809         final long analysis_start_time = new Date().getTime();
1810         PairwiseDomainSimilarityCalculator pw_calc = null;
1811         final DomainSimilarityCalculator calc = new BasicDomainSimilarityCalculator( domain_similarity_sort_field,
1812                                                                                      sort_by_species_count_first,
1813                                                                                      number_of_genomes == 2,
1814                                                                                      CALC_SIMILARITY_SCORES,
1815                                                                                      true );
1816         switch ( scoring ) {
1817             case COMBINATIONS:
1818                 pw_calc = new CombinationsBasedPairwiseDomainSimilarityCalculator();
1819                 break;
1820             case DOMAINS:
1821                 pw_calc = new DomainCountsBasedPairwiseSimilarityCalculator();
1822                 break;
1823             case PROTEINS:
1824                 pw_calc = new ProteinCountsBasedPairwiseDomainSimilarityCalculator();
1825                 break;
1826             default:
1827                 ForesterUtil.unexpectedFatalError( surfacing.PRG_NAME, "unknown value for sorting for scoring" );
1828         }
1829         DomainSimilarityCalculator.GoAnnotationOutput go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.NONE;
1830         if ( domain_id_to_go_ids_map != null ) {
1831             go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.ALL;
1832         }
1833         final SortedSet<DomainSimilarity> similarities = calc
1834                 .calculateSimilarities( pw_calc,
1835                                         gwcd_list,
1836                                         ignore_domains_without_combs_in_all_spec,
1837                                         ignore_species_specific_domains );
1838         SurfacingUtil.decoratePrintableDomainSimilarities( similarities, detailedness );
1839         final Map<String, Integer> tax_code_to_id_map = SurfacingUtil.createTaxCodeToIdMap( intrees[ 0 ] );
1840         try {
1841             String my_outfile = output_file.toString();
1842             Map<Character, Writer> split_writers = null;
1843             Writer writer = null;
1844             if ( similarities.size() > MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING ) {
1845                 if ( my_outfile.endsWith( ".html" ) ) {
1846                     my_outfile = my_outfile.substring( 0, my_outfile.length() - 5 );
1847                 }
1848                 split_writers = new HashMap<Character, Writer>();
1849                 SurfacingUtil.createSplitWriters( out_dir, my_outfile, split_writers );
1850             }
1851             else if ( !my_outfile.endsWith( ".html" ) ) {
1852                 my_outfile += ".html";
1853                 writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
1854             }
1855             List<Species> species_order = null;
1856             if ( species_matrix ) {
1857                 species_order = new ArrayList<Species>();
1858                 for( int i = 0; i < number_of_genomes; i++ ) {
1859                     species_order.add( new BasicSpecies( input_file_properties[ i ][ 1 ] ) );
1860                 }
1861             }
1862             html_desc.append( "<tr><td>Sum of all distinct binary combinations:</td><td>"
1863                     + all_bin_domain_combinations_encountered.size() + "</td></tr>" + nl );
1864             html_desc.append( "<tr><td>Sum of all distinct domains:</td><td>" + all_domains_encountered.size()
1865                               + "</td></tr>" + nl );
1866             html_desc.append( "<tr><td>Analysis date/time:</td><td>"
1867                     + new java.text.SimpleDateFormat( "yyyy.MM.dd HH:mm:ss" ).format( new java.util.Date() )
1868                     + "</td></tr>" + nl );
1869             html_desc.append( "</table>" + nl );
1870             final Writer simple_tab_writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR
1871                                                                                  + my_outfile + ".tsv" ) );
1872             SurfacingUtil.writeDomainSimilaritiesToFile( html_desc,
1873                                                          new StringBuilder( number_of_genomes + " genomes" ),
1874                                                          simple_tab_writer,
1875                                                          writer,
1876                                                          split_writers,
1877                                                          similarities,
1878                                                          number_of_genomes == 2,
1879                                                          species_order,
1880                                                          domain_similarity_print_option,
1881                                                          scoring,
1882                                                          true,
1883                                                          tax_code_to_id_map,
1884                                                          intree_0_orig,
1885                                                          positive_filter_file != null ? filter : null );
1886             simple_tab_writer.close();
1887             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote main output (includes domain similarities) to: \""
1888                     + ( out_dir == null ? my_outfile : out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) + "\"" );
1889         }
1890         catch ( final IOException e ) {
1891             ForesterUtil.fatalError( surfacing.PRG_NAME, "Failed to write similarites to: \"" + output_file + "\" ["
1892                     + e.getMessage() + "]" );
1893         }
1894         System.out.println();
1895         final Species[] species = new Species[ number_of_genomes ];
1896         for( int i = 0; i < number_of_genomes; ++i ) {
1897             species[ i ] = new BasicSpecies( input_file_properties[ i ][ 1 ] );
1898         }
1899         List<Phylogeny> inferred_trees = null;
1900         if ( ( number_of_genomes > 2 ) && perform_pwc ) {
1901             final PairwiseGenomeComparator pwgc = new PairwiseGenomeComparator();
1902             pwgc.performPairwiseComparisons( html_desc,
1903                                              sort_by_species_count_first,
1904                                              detailedness,
1905                                              ignore_domains_without_combs_in_all_spec,
1906                                              ignore_species_specific_domains,
1907                                              domain_similarity_sort_field_for_automated_pwc,
1908                                              domain_similarity_print_option,
1909                                              scoring,
1910                                              domain_id_to_go_ids_map,
1911                                              go_id_to_term_map,
1912                                              go_namespace_limit,
1913                                              species,
1914                                              number_of_genomes,
1915                                              gwcd_list,
1916                                              pw_calc,
1917                                              automated_pairwise_comparison_suffix,
1918                                              true,
1919                                              surfacing.PAIRWISE_DOMAIN_COMPARISONS_PREFIX,
1920                                              surfacing.PRG_NAME,
1921                                              out_dir,
1922                                              write_pwc_files,
1923                                              tax_code_to_id_map,
1924                                              CALC_SIMILARITY_SCORES,
1925                                              intree_0_orig );
1926             String matrix_output_file = new String( output_file.toString() );
1927             if ( matrix_output_file.indexOf( '.' ) > 1 ) {
1928                 matrix_output_file = matrix_output_file.substring( 0, matrix_output_file.indexOf( '.' ) );
1929             }
1930             if ( out_dir != null ) {
1931                 matrix_output_file = out_dir + ForesterUtil.FILE_SEPARATOR + matrix_output_file;
1932                 output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
1933             }
1934             SurfacingUtil.writeMatrixToFile( new File( matrix_output_file
1935                                                        + surfacing.MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ), pwgc.getDomainDistanceScoresMeans() );
1936             SurfacingUtil
1937             .writeMatrixToFile( new File( matrix_output_file
1938                                           + surfacing.MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ),
1939                                           pwgc.getSharedBinaryCombinationsBasedDistances() );
1940             SurfacingUtil.writeMatrixToFile( new File( matrix_output_file
1941                                                        + surfacing.MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ),
1942                                                        pwgc.getSharedDomainsBasedDistances() );
1943             final Phylogeny nj_gd = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
1944                                                                                              + surfacing.NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ), pwgc.getDomainDistanceScoresMeans()
1945                                                                                              .get( 0 ) );
1946             final Phylogeny nj_bc = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
1947                                                                                              + surfacing.NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
1948                                                                                              .getSharedBinaryCombinationsBasedDistances().get( 0 ) );
1949             final Phylogeny nj_d = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
1950                                                                                             + surfacing.NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
1951                                                                                             .getSharedDomainsBasedDistances().get( 0 ) );
1952             inferred_trees = new ArrayList<Phylogeny>();
1953             inferred_trees.add( nj_gd );
1954             inferred_trees.add( nj_bc );
1955             inferred_trees.add( nj_d );
1956         } // if ( ( output_file != null ) && ( number_of_genomes > 2 ) && !isEmpty( automated_pairwise_comparison_suffix ) )
1957         if ( ( out_dir != null ) && ( !perform_pwc ) ) {
1958             output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
1959         }
1960         if ( write_to_nexus ) {
1961             SurfacingUtil.writePresentToNexus( output_file, positive_filter_file, filter, gwcd_list );
1962         }
1963         if ( ( ( intrees != null ) && ( intrees.length > 0 ) ) && ( number_of_genomes > 2 ) ) {
1964             final StringBuilder parameters_sb = SurfacingUtil.createParametersAsString( ignore_dufs,
1965                                                                                         ie_value_max,
1966                                                                                         fs_e_value_max,
1967                                                                                         max_allowed_overlap,
1968                                                                                         no_engulfing_overlaps,
1969                                                                                         cutoff_scores_file,
1970                                                                                         dc_type );
1971             String s = "_";
1972             if ( radomize_fitch_parsimony ) {
1973                 s += random_number_seed_for_fitch_parsimony + "_";
1974             }
1975             int i = 0;
1976             for( final Phylogeny intree : intrees ) {
1977                 final String outfile_name = ForesterUtil.removeSuffix( output_file.toString() ) + s
1978                         + ForesterUtil.removeSuffix( intree_files[ i ].toString() );
1979                 final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator.createInstance( intree,
1980                                                                                                              gwcd_list );
1981                 SurfacingUtil.executeParsimonyAnalysis( random_number_seed_for_fitch_parsimony,
1982                                                         radomize_fitch_parsimony,
1983                                                         outfile_name,
1984                                                         domain_parsimony,
1985                                                         intree,
1986                                                         domain_id_to_go_ids_map,
1987                                                         go_id_to_term_map,
1988                                                         go_namespace_limit,
1989                                                         parameters_sb.toString(),
1990                                                         domain_id_to_secondary_features_maps,
1991                                                         positive_filter_file == null ? null : filter,
1992                                                                 output_binary_domain_combinationsfor_graph_analysis,
1993                                                                 all_bin_domain_combinations_gained_fitch,
1994                                                                 all_bin_domain_combinations_lost_fitch,
1995                                                                 dc_type,
1996                                                                 protein_length_stats_by_dc,
1997                                                                 domain_number_stats_by_dc,
1998                                                                 domain_length_stats_by_domain,
1999                                                                 tax_code_to_id_map,
2000                                                                 write_to_nexus,
2001                                                                 use_last_in_fitch_parsimony,
2002                                                                 perform_dc_fich );
2003                 // Listing of all domain combinations gained is only done if only one input tree is used.
2004                 if ( ( domain_id_to_secondary_features_maps != null )
2005                         && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
2006                     int j = 0;
2007                     for( final Map<String, Set<String>> domain_id_to_secondary_features_map : domain_id_to_secondary_features_maps ) {
2008                         final Map<Species, MappingResults> mapping_results_map = new TreeMap<Species, MappingResults>();
2009                         final DomainParsimonyCalculator secondary_features_parsimony = DomainParsimonyCalculator
2010                                 .createInstance( intree, gwcd_list, domain_id_to_secondary_features_map );
2011                         SurfacingUtil
2012                         .executeParsimonyAnalysisForSecondaryFeatures( outfile_name
2013                                                                        + "_"
2014                                                                        + secondary_features_map_files[ j++ ],
2015                                                                        secondary_features_parsimony,
2016                                                                        intree,
2017                                                                        parameters_sb.toString(),
2018                                                                        mapping_results_map,
2019                                                                        use_last_in_fitch_parsimony );
2020                         if ( i == 0 ) {
2021                             System.out.println();
2022                             System.out.println( "Mapping to secondary features:" );
2023                             for( final Species spec : mapping_results_map.keySet() ) {
2024                                 final MappingResults mapping_results = mapping_results_map.get( spec );
2025                                 final int total_domains = mapping_results.getSumOfFailures()
2026                                         + mapping_results.getSumOfSuccesses();
2027                                 System.out.print( spec + ":" );
2028                                 System.out.print( " mapped domains = " + mapping_results.getSumOfSuccesses() );
2029                                 System.out.print( ", not mapped domains = " + mapping_results.getSumOfFailures() );
2030                                 if ( total_domains > 0 ) {
2031                                     System.out.println( ", mapped ratio = "
2032                                             + ( ( 100 * mapping_results.getSumOfSuccesses() ) / total_domains ) + "%" );
2033                                 }
2034                                 else {
2035                                     System.out.println( ", mapped ratio = n/a (total domains = 0 )" );
2036                                 }
2037                             }
2038                         }
2039                     }
2040                 }
2041                 i++;
2042             } // for( final Phylogeny intree : intrees ) {
2043         }
2044         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
2045             SurfacingUtil.executePlusMinusAnalysis( output_file,
2046                                                     plus_minus_analysis_high_copy_base_species,
2047                                                     plus_minus_analysis_high_copy_target_species,
2048                                                     plus_minus_analysis_high_low_copy_species,
2049                                                     gwcd_list,
2050                                                     protein_lists_per_species,
2051                                                     domain_id_to_go_ids_map,
2052                                                     go_id_to_term_map,
2053                                                     plus_minus_analysis_numbers );
2054         }
2055         if ( output_protein_lists_for_all_domains ) {
2056             SurfacingUtil.writeProteinListsForAllSpecies( out_dir,
2057                                                           protein_lists_per_species,
2058                                                           gwcd_list,
2059                                                           output_list_of_all_proteins_per_domain_e_value_max,
2060                                                           positive_filter_file != null ? filter : null );
2061         }
2062         gwcd_list = null;
2063         if ( all_bin_domain_combinations_gained_fitch != null ) {
2064             try {
2065                 SurfacingUtil
2066                 .executeFitchGainsAnalysis( new File( output_file
2067                                                       + surfacing.OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2068                                                       all_bin_domain_combinations_gained_fitch,
2069                                                       all_domains_encountered.size(),
2070                                                       all_bin_domain_combinations_encountered,
2071                                                       true );
2072             }
2073             catch ( final IOException e ) {
2074                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2075             }
2076         }
2077         if ( all_bin_domain_combinations_lost_fitch != null ) {
2078             try {
2079                 SurfacingUtil
2080                 .executeFitchGainsAnalysis( new File( output_file
2081                                                       + surfacing.OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2082                                                       all_bin_domain_combinations_lost_fitch,
2083                                                       all_domains_encountered.size(),
2084                                                       all_bin_domain_combinations_encountered,
2085                                                       false );
2086             }
2087             catch ( final IOException e ) {
2088                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2089             }
2090         }
2091         final Runtime rt = java.lang.Runtime.getRuntime();
2092         final long free_memory = rt.freeMemory() / 1000000;
2093         final long total_memory = rt.totalMemory() / 1000000;
2094         ForesterUtil.programMessage( PRG_NAME, "Time for analysis : " + ( new Date().getTime() - analysis_start_time )
2095                                      + "ms" );
2096         ForesterUtil.programMessage( PRG_NAME, "Total running time: " + ( new Date().getTime() - start_time ) + "ms " );
2097         ForesterUtil.programMessage( PRG_NAME, "Free memory       : " + free_memory + "MB, total memory: "
2098                 + total_memory + "MB" );
2099         ForesterUtil.programMessage( PRG_NAME, "If this application is useful to you, please cite:" );
2100         ForesterUtil.programMessage( PRG_NAME, surfacing.WWW );
2101         ForesterUtil.programMessage( PRG_NAME, "OK" );
2102         System.out.println();
2103     }
2104
2105     private static void printHelp() {
2106         System.out.println();
2107         System.out.println( "Usage:" );
2108         System.out.println();
2109         System.out.println( "% java -Xms256m -Xmx512m -cp forester.jar org.forester.applications." + surfacing.PRG_NAME
2110                             + " [options] <phylogen(y|ies) infile> [external node name 1] [name 2] ... [name n]" );
2111         System.out.println();
2112         System.out.println( " Note: This software might need a significant amount of memory (heap space);" );
2113         System.out
2114         .println( "       hence use \"-Xms128m -Xmx512m\" (or more) to prevent a \"java.lang.OutOfMemoryError\"." );
2115         System.out.println();
2116         System.out.println( " Options: " );
2117         System.out.println( surfacing.DETAILEDNESS_OPTION + ": level of detail for similarities output file (default:"
2118                 + DETAILEDNESS_DEFAULT + ")" );
2119         System.out.println( surfacing.IGNORE_COMBINATION_WITH_SAME_OPTION
2120                             + ": to ignore combinations with self (default: not to ignore)" );
2121         System.out
2122         .println( surfacing.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION
2123                   + ": to ignore domains without combinations in any species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2124         System.out
2125         .println( surfacing.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION
2126                   + ": to ignore domains specific to one species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2127         System.out.println( surfacing.NOT_IGNORE_DUFS_OPTION
2128                             + ": to _not_ ignore DUFs (domains with unknown function) (default: ignore DUFs)" );
2129         System.out
2130         .println( surfacing.IGNORE_VIRAL_IDS
2131                   + ": to ignore domains with ids containing 'vir', 'retro', 'transpos', 'phage', or starting with 'rv' or 'gag_'" );
2132         System.out.println( surfacing.DOMAIN_SIMILARITY_SORT_OPTION + ": sorting for similarities (default: "
2133                 + DOMAIN_SORT_FILD_DEFAULT + ")" );
2134         System.out.println( surfacing.OUTPUT_FILE_OPTION + ": name for (main) output file (mandatory)" );
2135         System.out.println( surfacing.MAX_I_E_VALUE_OPTION + ": max (inclusive) iE-value" );
2136         System.out.println( surfacing.MAX_FS_E_VALUE_OPTION + ": max (inclusive) FS E-value" );
2137         System.out.println( surfacing.MAX_ALLOWED_OVERLAP_OPTION + ": maximal allowed domain overlap" );
2138         System.out.println( surfacing.NO_ENGULFING_OVERLAP_OPTION + ": to ignore engulfed lower confidence domains" );
2139         System.out.println( surfacing.SPECIES_MATRIX_OPTION + ": species matrix" );
2140         System.out.println( surfacing.SCORING_OPTION + ": scoring (default:" + SCORING_DEFAULT + ")" );
2141         System.out.println( surfacing.DOMAIN_COUNT_SORT_OPTION + ": sorting for domain counts (default:"
2142                 + DOMAINS_SORT_ORDER_DEFAULT + ")" );
2143         System.out.println( surfacing.DOMAIN_SIMILARITY_PRINT_OPTION + ": domain similarity print option (default:"
2144                 + DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT + ")" );
2145         System.out.println( surfacing.CUTOFF_SCORE_FILE_OPTION + ": cutoff score file" );
2146         System.out.println( surfacing.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION
2147                             + ": sort by species count first" );
2148         System.out.println( surfacing.OUTPUT_DIR_OPTION + ": output directory" );
2149         System.out.println( surfacing.PFAM_TO_GO_FILE_USE_OPTION + ": Pfam to GO mapping file" );
2150         System.out.println( surfacing.GO_OBO_FILE_USE_OPTION + ": GO terms file (OBO format)" );
2151         System.out.println( surfacing.GO_NAMESPACE_LIMIT_OPTION + ": limit GO term to one GO namespace" );
2152         System.out.println( surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
2153                             + "[=<suffix for pairwise comparison output files>]: to perform pairwise comparison based analyses" );
2154         System.out.println( surfacing.INPUT_SPECIES_TREE_OPTION
2155                             + ": species tree, to perform (Dollo, Fitch) parismony analyses" );
2156         System.out
2157         .println( surfacing.INPUT_SPECIES_TREE_OPTION
2158                   + "=<treefiles in phyloXML format, separated by #>: to infer domain/binary domain combination gains/losses on given species trees" );
2159         System.out.println( surfacing.FILTER_POSITIVE_OPTION
2160                             + "=<file>: to filter out proteins not containing at least one domain listed in <file>" );
2161         System.out.println( surfacing.FILTER_NEGATIVE_OPTION
2162                             + "=<file>: to filter out proteins containing at least one domain listed in <file>" );
2163         System.out.println( surfacing.FILTER_NEGATIVE_DOMAINS_OPTION
2164                             + "=<file>: to filter out (ignore) domains listed in <file>" );
2165         System.out.println( surfacing.INPUT_GENOMES_FILE_OPTION + "=<file>: to read input files from <file>" );
2166         System.out
2167         .println( surfacing.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION
2168                   + "=<seed>: seed for random number generator for Fitch Parsimony analysis (type: long, default: no randomization - given a choice, prefer absence" );
2169         System.out.println( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS
2170                             + ": to consider directedness in binary combinations: e.g. A-B != B-A" );
2171         System.out.println( surfacing.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY
2172                             + ": to consider directedness and adjacency in binary combinations" );
2173         System.out
2174         .println( surfacing.SEQ_EXTRACT_OPTION
2175                   + "=<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')" );
2176         System.out.println( surfacing.SECONDARY_FEATURES_PARSIMONY_MAP_FILE
2177                             + "=<file>: to perfom parsimony analysis on secondary features" );
2178         System.out.println( surfacing.PLUS_MINUS_ANALYSIS_OPTION + "=<file>: to presence/absence genome analysis" );
2179         System.out.println( surfacing.DOMAIN_COMBINITONS_COUNTS_OUTPUT_OPTION
2180                             + ": to output binary domain counts (as individual files)" );
2181         System.out.println( surfacing.DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS
2182                             + ": to output binary domain combinations for (downstream) graph analysis" );
2183         System.out.println( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS + ": to output all proteins per domain" );
2184         System.out.println( surfacing.OUTPUT_LIST_OF_ALL_PROTEINS_PER_DOMAIN_E_VALUE_OPTION
2185                             + ": e value max per domain for output of all proteins per domain" );
2186         System.out.println( surfacing.USE_LAST_IN_FITCH_OPTION + ": to use last in Fitch parsimony" );
2187         System.out.println( surfacing.WRITE_TO_NEXUS_OPTION + ": to output in Nexus format" );
2188         System.out.println( PERFORM_DC_FITCH + ": to perform DC Fitch parsimony" );
2189         System.out.println( PERFORM_DC_REGAIN_PROTEINS_STATS_OPTION + ": to perform DC regain protein statistics" );
2190         System.out.println( DA_ANALYSIS_OPTION + ": to perform DA analysis" );
2191         System.out.println( PERFORM_DOMAIN_LENGTH_ANALYSIS_OPTION + ": to perform domain length analysis" );
2192         System.out.println();
2193         System.out.println();
2194         System.out
2195         .println( "Example 1: surfacing -p2g=pfam2go.txt -obo=go.obo -species_tree=tol_156.xml -no_eo -ie=0.01 -dufs -genomes=genomes_all.txt -pos_filter=tf_1.txt -out_dir=_tf1 -o=tf1" );
2196         System.out.println();
2197         System.out
2198         .println( "Example 2: surfacing -p2g=pfam2go.txt -obo=go.obo -species_tree=tol_156.xml -last -ignore_viral_ids -no_eo -ie=0.1 -dufs -genomes=genomes_all.txt -pos_filter=tf_1.txt -all_prot -all_prot_e=0.1 -out_dir=_tf1_e01_ape01 -o=tf1_e01_ape01" );
2199         System.out.println();
2200     }
2201 }