forester/java/src/org/forester/application/surfacing_hmmpfam.java

   1 // $Id:
   2 //
   3 // FORESTER -- software libraries and applications
   4 // for evolutionary biology research and applications.
   5 //
   6 // Copyright (C) 2008-2009 Christian M. Zmasek
   7 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   8 // All rights reserved
   9 //
  10 // This library is free software; you can redistribute it and/or
  11 // modify it under the terms of the GNU Lesser General Public
  12 // License as published by the Free Software Foundation; either
  13 // version 2.1 of the License, or (at your option) any later version.
  14 //
  15 // This library is distributed in the hope that it will be useful,
  16 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18 // Lesser General Public License for more details.
  19 //
  20 // You should have received a copy of the GNU Lesser General Public
  21 // License along with this library; if not, write to the Free Software
  22 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  23 //
  24 // Contact: phylosoft @ gmail . com
  25 // WWW: www.phylosoft.org/forester
  26
  27 package org.forester.application;
  28
  29 import java.io.BufferedWriter;
  30 import java.io.File;
  31 import java.io.FileWriter;
  32 import java.io.IOException;
  33 import java.io.Writer;
  34 import java.util.ArrayList;
  35 import java.util.Date;
  36 import java.util.HashSet;
  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.evoinference.distance.NeighborJoining;
  46 import org.forester.evoinference.matrix.distance.DistanceMatrix;
  47 import org.forester.go.GoId;
  48 import org.forester.go.GoNameSpace;
  49 import org.forester.go.GoTerm;
  50 import org.forester.go.GoUtils;
  51 import org.forester.go.OBOparser;
  52 import org.forester.go.PfamToGoMapping;
  53 import org.forester.go.PfamToGoParser;
  54 import org.forester.io.parsers.HmmPfamOutputParser;
  55 import org.forester.io.parsers.util.ParserUtils;
  56 import org.forester.io.writers.PhylogenyWriter;
  57 import org.forester.phylogeny.Phylogeny;
  58 import org.forester.phylogeny.PhylogenyMethods;
  59 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  60 import org.forester.surfacing.BasicDomainSimilarityCalculator;
  61 import org.forester.surfacing.BasicGenomeWideCombinableDomains;
  62 import org.forester.surfacing.BasicSpecies;
  63 import org.forester.surfacing.BinaryDomainCombination;
  64 import org.forester.surfacing.CombinationsBasedPairwiseDomainSimilarityCalculator;
  65 import org.forester.surfacing.DomainCountsBasedPairwiseSimilarityCalculator;
  66 import org.forester.surfacing.DomainCountsDifferenceUtil;
  67 import org.forester.surfacing.DomainId;
  68 import org.forester.surfacing.DomainLengthsTable;
  69 import org.forester.surfacing.DomainParsimonyCalculator;
  70 import org.forester.surfacing.DomainSimilarity;
  71 import org.forester.surfacing.DomainSimilarity.DomainSimilarityScoring;
  72 import org.forester.surfacing.DomainSimilarity.DomainSimilaritySortField;
  73 import org.forester.surfacing.DomainSimilarityCalculator;
  74 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  75 import org.forester.surfacing.GenomeWideCombinableDomains;
  76 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  77 import org.forester.surfacing.MappingResults;
  78 import org.forester.surfacing.PairwiseDomainSimilarityCalculator;
  79 import org.forester.surfacing.PairwiseGenomeComparator;
  80 import org.forester.surfacing.PrintableDomainSimilarity;
  81 import org.forester.surfacing.PrintableDomainSimilarity.PRINT_OPTION;
  82 import org.forester.surfacing.Protein;
  83 import org.forester.surfacing.ProteinCountsBasedPairwiseDomainSimilarityCalculator;
  84 import org.forester.surfacing.Species;
  85 import org.forester.surfacing.SurfacingUtil;
  86 import org.forester.util.BasicTable;
  87 import org.forester.util.BasicTableParser;
  88 import org.forester.util.CommandLineArguments;
  89 import org.forester.util.DescriptiveStatistics;
  90 import org.forester.util.ForesterConstants;
  91 import org.forester.util.ForesterUtil;
  92
  93 public class surfacing_hmmpfam {
  94
  95     public final static String                                DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS                    = "graph_analysis_out";
  96     public final static String                                DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS                = "_dc.dot";
  97     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_fitch_present_dc.dot";
  98     public final static String                                DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX                             = ".dcc";
  99     // gain/loss:
 100     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS                               = "_dollo_gl_d";
 101     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_BINARY_COMBINATIONS                   = "_dollo_gl_dc";
 102     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_DOMAINS                               = "_fitch_gl_d";
 103     public final static String                                PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS                   = "_fitch_gl_dc";
 104     // gain/loss counts:
 105     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS                        = "_dollo_glc_d";
 106     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_BINARY_COMBINATIONS            = "_dollo_glc_dc";
 107     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_DOMAINS                        = "_fitch_glc_d";
 108     public final static String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS            = "_fitch_glc_dc";
 109     // tables:
 110     public final static String                                PARSIMONY_OUTPUT_FITCH_GAINS_BC                                        = "_fitch_gains_dc";
 111     public final static String                                PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC                                   = "_fitch_gains_dc.html";
 112     public final static String                                PARSIMONY_OUTPUT_FITCH_LOSSES_BC                                       = "_fitch_losses_dc";
 113     public final static String                                PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC                                  = "_fitch_losses_dc.html";
 114     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_BC                                      = "_fitch_present_dc";
 115     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC                                 = "_fitch_present_dc.html";
 116     public final static String                                PARSIMONY_OUTPUT_DOLLO_GAINS_D                                         = "_dollo_gains_d";
 117     public final static String                                PARSIMONY_OUTPUT_DOLLO_GAINS_GOID_D                                    = "_dollo_gains_goid_d";
 118     public final static String                                PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D                                    = "_dollo_gains_d.html";
 119     public final static String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_D                                        = "_dollo_losses_d";
 120     public final static String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D                                   = "_dollo_losses_d.html";
 121     public final static String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_D                                       = "_dollo_present_d";
 122     public final static String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_GOID_D                                  = "_dollo_present_goid_d";
 123     public final static String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D                                  = "_dollo_present_d.html";
 124     public final static String                                DOMAINS_PRESENT_NEXUS                                                  = "_dom.nex";
 125     public final static String                                BDC_PRESENT_NEXUS                                                      = "_dc.nex";
 126     // ---
 127     public final static String                                PRG_NAME                                                               = "surfacing";
 128     public static final String                                DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                             = "_d_dollo"
 129                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 130     public static final String                                DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH                             = "_d_fitch"
 131                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 132     public static final String                                BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO          = "_dc_dollo"
 133                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 134     public static final String                                BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH          = "_dc_fitch"
 135                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 136     public static final String                                NEXUS_EXTERNAL_DOMAINS                                                 = "_dom.nex";
 137     public static final String                                NEXUS_EXTERNAL_DOMAIN_COMBINATIONS                                     = "_dc.nex";
 138     public static final String                                NEXUS_SECONDARY_FEATURES                                               = "_secondary_features.nex";
 139     public static final String                                PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES                    = "_dollo_gl_secondary_features";
 140     public static final String                                PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES             = "_dollo_glc_secondary_features";
 141     public static final String                                PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES                        = "_dollo_gains_secondary_features";
 142     public static final String                                PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES                       = "_dollo_losses_secondary_features";
 143     public static final String                                PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES                      = "_dollo_present_secondary_features";
 144     public static final String                                SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO                  = "_secondary_features_dollo"
 145                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 146     public static final String                                PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_BIOLOGICAL_PROCESS                   = "_dollo_biol_proc_goid_d";
 147     public static final String                                PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_CELLULAR_COMPONENT                   = "_dollo_cell_comp_goid_d";
 148     public static final String                                PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_MOLECULAR_FUNCTION                   = "_dollo_mol_funct_goid_d";
 149     public static final String                                PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_ALL_NAMESPACES                       = "_dollo_goid_d";
 150     public static final String                                PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_BIOLOGICAL_PROCESS                  = "_fitch_biol_proc_goid_dc";
 151     public static final String                                PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_CELLULAR_COMPONENT                  = "_fitch_cell_comp_goid_dc";
 152     public static final String                                PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_MOLECULAR_FUNCTION                  = "_fitch_mol_funct_goid_dc";
 153     public static final String                                PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_ALL_NAMESPACES                      = "_fitch_goid_dc";
 154     final static private String                               HELP_OPTION_1                                                          = "help";
 155     final static private String                               HELP_OPTION_2                                                          = "h";
 156     final static private String                               OUTPUT_DIR_OPTION                                                      = "out_dir";
 157     final static private String                               SCORING_OPTION                                                         = "scoring";
 158     private static final DomainSimilarityScoring              SCORING_DEFAULT                                                        = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 159     final static private String                               SCORING_DOMAIN_COUNT_BASED                                             = "domains";
 160     final static private String                               SCORING_PROTEIN_COUNT_BASED                                            = "proteins";
 161     final static private String                               SCORING_COMBINATION_BASED                                              = "combinations";
 162     final static private String                               DETAILEDNESS_OPTION                                                    = "detail";
 163     private final static Detailedness                         DETAILEDNESS_DEFAULT                                                   = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 164     final static private String                               SPECIES_MATRIX_OPTION                                                  = "smatrix";
 165     final static private String                               DETAILEDNESS_BASIC                                                     = "basic";
 166     final static private String                               DETAILEDNESS_LIST_IDS                                                  = "list_ids";
 167     final static private String                               DETAILEDNESS_PUNCTILIOUS                                               = "punctilious";
 168     final static private String                               DOMAIN_SIMILARITY_SORT_OPTION                                          = "sort";
 169     private static final DomainSimilaritySortField            DOMAIN_SORT_FILD_DEFAULT                                               = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 170     final static private String                               DOMAIN_SIMILARITY_SORT_MIN                                             = "min";
 171     final static private String                               DOMAIN_SIMILARITY_SORT_MAX                                             = "max";
 172     final static private String                               DOMAIN_SIMILARITY_SORT_SD                                              = "sd";
 173     final static private String                               DOMAIN_SIMILARITY_SORT_MEAN                                            = "mean";
 174     final static private String                               DOMAIN_SIMILARITY_SORT_DIFF                                            = "diff";
 175     final static private String                               DOMAIN_SIMILARITY_SORT_COUNTS_DIFF                                     = "count_diff";
 176     final static private String                               DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF                                 = "abs_count_diff";
 177     final static private String                               DOMAIN_SIMILARITY_SORT_SPECIES_COUNT                                   = "species";
 178     final static private String                               DOMAIN_SIMILARITY_SORT_ALPHA                                           = "alpha";
 179     final static private String                               DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION                   = "species_first";
 180     final static private String                               DOMAIN_COUNT_SORT_OPTION                                               = "dc_sort";
 181     private static final GenomeWideCombinableDomainsSortOrder DOMAINS_SORT_ORDER_DEFAULT                                             = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
 182     final static private String                               DOMAIN_COUNT_SORT_ALPHA                                                = "alpha";
 183     final static private String                               DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT                                     = "dom";
 184     final static private String                               DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT                            = "prot";
 185     final static private String                               DOMAIN_COUNT_SORT_COMBINATIONS_COUNT                                   = "comb";
 186     final static private String                               CUTOFF_SCORE_FILE_OPTION                                               = "cos";
 187     final static private String                               NOT_IGNORE_DUFS_OPTION                                                 = "dufs";
 188     final static private String                               MAX_E_VALUE_OPTION                                                     = "e";
 189     final static private String                               MAX_ALLOWED_OVERLAP_OPTION                                             = "mo";
 190     final static private String                               NO_ENGULFING_OVERLAP_OPTION                                            = "no_eo";
 191     final static private String                               IGNORE_COMBINATION_WITH_SAME_OPTION                                    = "ignore_self_comb";
 192     final static private String                               PAIRWISE_DOMAIN_COMPARISONS_PREFIX                                     = "pwc_";
 193     final static private String                               PAIRWISE_DOMAIN_COMPARISONS_OPTION                                     = "pwc";
 194     final static private String                               OUTPUT_FILE_OPTION                                                     = "o";
 195     final static private String                               PFAM_TO_GO_FILE_USE_OPTION                                             = "p2g";
 196     final static private String                               GO_OBO_FILE_USE_OPTION                                                 = "obo";
 197     final static private String                               GO_NAMESPACE_LIMIT_OPTION                                              = "go_namespace";
 198     final static private String                               GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION                           = "molecular_function";
 199     final static private String                               GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS                           = "biological_process";
 200     final static private String                               GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT                           = "cellular_component";
 201     final static private String                               SECONDARY_FEATURES_PARSIMONY_MAP_FILE                                  = "secondary";
 202     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED                    = "simple_tab";
 203     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML                             = "simple_html";
 204     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML                           = "detailed_html";
 205     final static private String                               DOMAIN_SIMILARITY_PRINT_OPTION                                         = "ds_output";
 206     private static final PRINT_OPTION                         DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT                                 = PrintableDomainSimilarity.PRINT_OPTION.HTML;
 207     final static private String                               IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION              = "ignore_singlet_domains";
 208     final static private String                               IGNORE_VIRAL_IDS                                                       = "ignore_viral_ids";
 209     final static private boolean                              IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT             = false;
 210     final static private String                               IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION                          = "ignore_species_specific_domains";
 211     final static private boolean                              IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT                  = false;
 212     final static private String                               MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                         = "_mean_score.pwd";
 213     final static private String                               MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                     = "_domains.pwd";
 214     final static private String                               MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX            = "_bin_combinations.pwd";
 215     final static private String                               NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX                        = "_mean_score_NJ"
 216                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 217     final static private String                               NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX                    = "_domains_NJ"
 218                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 219     final static private String                               NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX           = "_bin_combinations_NJ"
 220                                                                                                                                              + ForesterConstants.PHYLO_XML_SUFFIX;
 221     final static private String                               DISPLAY_M_HISTOGRAMS_OPTION                                            = "mhisto";
 222     //  final static private boolean DISPLAY_M_HISTOGRAMS_OPTION_DEFAULT                                    = false;
 223     final static private String                               JACKNIFE_OPTION                                                        = "jack";
 224     final static private String                               JACKNIFE_RANDOM_SEED_OPTION                                            = "seed";
 225     final static private String                               JACKNIFE_RATIO_OPTION                                                  = "jack_ratio";
 226     private static final int                                  JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT                                 = 100;
 227     final static private long                                 JACKNIFE_RANDOM_SEED_DEFAULT                                           = 19;
 228     final static private double                               JACKNIFE_RATIO_DEFAULT                                                 = 0.5;
 229     //final static private String  INFER_SPECIES_TREES_OPTION                                             = "species_tree_inference";
 230     final static private String                               INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX                               = "_sd_nj.nh";
 231     final static private String                               INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX                              = "_sbc_nj.nh";
 232     final static private String                               FILTER_POSITIVE_OPTION                                                 = "pos_filter";
 233     final static private String                               FILTER_NEGATIVE_OPTION                                                 = "neg_filter";
 234     final static private String                               FILTER_NEGATIVE_DOMAINS_OPTION                                         = "neg_dom_filter";
 235     final static private String                               INPUT_FILES_FROM_FILE_OPTION                                           = "input";
 236     final static private String                               INPUT_SPECIES_TREE_OPTION                                              = "species_tree";
 237     final static private String                               SEQ_EXTRACT_OPTION                                                     = "prot_extract";
 238     final static private char                                 SEPARATOR_FOR_INPUT_VALUES                                             = '#';
 239     final static private String                               PRG_VERSION                                                            = "1.00";
 240     final static private String                               PRG_DATE                                                               = "2009.07.06";
 241     final static private String                               E_MAIL                                                                 = "czmasek@burnham.org";
 242     final static private String                               WWW                                                                    = "www.phylosoft.org/forester/applications/surfacing";
 243     final static private boolean                              IGNORE_DUFS_DEFAULT                                                    = true;
 244     final static private boolean                              IGNORE_COMBINATION_WITH_SAME_DEFAULLT                                  = false;
 245     final static private double                               MAX_E_VALUE_DEFAULT                                                    = -1;
 246     final static private int                                  MAX_ALLOWED_OVERLAP_DEFAULT                                            = -1;
 247     final static private String                               DEFAULT_SEARCH_PARAMETER                                               = "ls";
 248     final private static boolean                              ALLOW_NON_UNIQUE_QUERY_IN_HMMPFAM_OUTPUT_DEFAULT                       = true;
 249     final private static boolean                              VERBOSE_DEFAULT                                                        = true;
 250     private static final String                               RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION                                 = "random_seed";
 251     private static final String                               CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS                               = "consider_bdc_direction";
 252     private static final String                               CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY                 = "consider_bdc_adj";
 253     private static final String                               SEQ_EXTRACT_SUFFIX                                                     = ".prot";
 254     private static final String                               PLUS_MINUS_ANALYSIS_OPTION                                             = "plus_minus";
 255     private static final String                               PLUS_MINUS_DOM_SUFFIX                                                  = "_plus_minus_dom.txt";
 256     private static final String                               PLUS_MINUS_DOM_SUFFIX_HTML                                             = "_plus_minus_dom.html";
 257     private static final String                               PLUS_MINUS_DC_SUFFIX_HTML                                              = "_plus_minus_dc.html";
 258     private static final int                                  PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT                                   = 0;
 259     private static final double                               PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT                                     = 1.0;
 260     private static final String                               PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX                                       = "_plus_minus_go_ids_all.txt";
 261     private static final String                               PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX                                   = "_plus_minus_go_ids_passing.txt";
 262     private static final String                               OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS                                    = "all_prot";
 263     private static final boolean                              VERBOSE                                                                = false;
 264     private static final String                               OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX       = "_fitch_dc_gains_counts";
 265     private static final String                               OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX         = "_fitch_dc_losses_counts";
 266     private static final String                               DOMAIN_LENGTHS_ANALYSIS_SUFFIX                                         = "_domain_lengths_analysis";
 267     private static final boolean                              PERFORM_DOMAIN_LENGTH_ANALYSIS                                         = true;
 268     public static final String                                ALL_PFAMS_ENCOUNTERED_SUFFIX                                           = "_all_encountered_pfams";
 269     public static final String                                ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX                        = "_all_encountered_pfams_with_go_annotation";
 270     public static final String                                ENCOUNTERED_PFAMS_SUMMARY_SUFFIX                                       = "_encountered_pfams_summary";
 271     public static final String                                ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX                                     = "_all_pfams_gained_as_domains";
 272     public static final String                                ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX                                       = "_all_pfams_lost_as_domains";
 273     public static final String                                ALL_PFAMS_GAINED_AS_DC_SUFFIX                                          = "_all_pfams_gained_as_dc";
 274     public static final String                                ALL_PFAMS_LOST_AS_DC_SUFFIX                                            = "_all_pfams_lost_as_dc";
 275     public static final String                                BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES                         = "PER_NODE_EVENTS";
 276     public static final String                                BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES                      = "PER_SUBTREE_EVENTS";
 277     public static final String                                D_PROMISCUITY_FILE_SUFFIX                                              = "_domain_promiscuities";
 278
 279     // final String error = ForesterUtil.isReadableFile( new File(
 280     // input_file_properties[ i ][ 0 ] ) );
 281     // if ( !ForesterUtil.isEmpty( error ) ) {
 282     // ForesterUtil.fatalError( surfacing.PRG_NAME, error );
 283     // }
 284     private static void checkWriteabilityForPairwiseComparisons( final PrintableDomainSimilarity.PRINT_OPTION domain_similarity_print_option,
 285                                                                  final String[][] input_file_properties,
 286                                                                  final String automated_pairwise_comparison_suffix,
 287                                                                  final File outdir ) {
 288         for( int i = 0; i < input_file_properties.length; ++i ) {
 289             for( int j = 0; j < i; ++j ) {
 290                 final String species_i = input_file_properties[ i ][ 1 ];
 291                 final String species_j = input_file_properties[ j ][ 1 ];
 292                 String pairwise_similarities_output_file_str = PAIRWISE_DOMAIN_COMPARISONS_PREFIX + species_i + "_"
 293                         + species_j + automated_pairwise_comparison_suffix;
 294                 switch ( domain_similarity_print_option ) {
 295                     case HTML:
 296                         if ( !pairwise_similarities_output_file_str.endsWith( ".html" ) ) {
 297                             pairwise_similarities_output_file_str += ".html";
 298                         }
 299                         break;
 300                 }
 301                 final String error = ForesterUtil
 302                         .isWritableFile( new File( outdir == null ? pairwise_similarities_output_file_str : outdir
 303                                 + ForesterUtil.FILE_SEPARATOR + pairwise_similarities_output_file_str ) );
 304                 if ( !ForesterUtil.isEmpty( error ) ) {
 305                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, error );
 306                 }
 307             }
 308         }
 309     }
 310
 311     private static StringBuilder createParametersAsString( final boolean ignore_dufs,
 312                                                            final double e_value_max,
 313                                                            final int max_allowed_overlap,
 314                                                            final boolean no_engulfing_overlaps,
 315                                                            final File cutoff_scores_file,
 316                                                            final BinaryDomainCombination.DomainCombinationType dc_type ) {
 317         final StringBuilder parameters_sb = new StringBuilder();
 318         parameters_sb.append( "E-value: " + e_value_max );
 319         if ( cutoff_scores_file != null ) {
 320             parameters_sb.append( ", Cutoff-scores-file: " + cutoff_scores_file );
 321         }
 322         else {
 323             parameters_sb.append( ", Cutoff-scores-file: not-set" );
 324         }
 325         if ( max_allowed_overlap != surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_DEFAULT ) {
 326             parameters_sb.append( ", Max-overlap: " + max_allowed_overlap );
 327         }
 328         else {
 329             parameters_sb.append( ", Max-overlap: not-set" );
 330         }
 331         if ( no_engulfing_overlaps ) {
 332             parameters_sb.append( ", Engulfing-overlaps: not-allowed" );
 333         }
 334         else {
 335             parameters_sb.append( ", Engulfing-overlaps: allowed" );
 336         }
 337         if ( ignore_dufs ) {
 338             parameters_sb.append( ", Ignore-dufs: true" );
 339         }
 340         else {
 341             parameters_sb.append( ", Ignore-dufs: false" );
 342         }
 343         parameters_sb.append( ", DC type (if applicable): " + dc_type );
 344         return parameters_sb;
 345     }
 346
 347     /**
 348      * Warning: This sideeffects 'all_bin_domain_combinations_encountered'!
 349      *
 350      *
 351      * @param output_file
 352      * @param all_bin_domain_combinations_changed
 353      * @param sum_of_all_domains_encountered
 354      * @param all_bin_domain_combinations_encountered
 355      * @param is_gains_analysis
 356      * @throws IOException
 357      */
 358     private static void executeFitchGainsAnalysis( final File output_file,
 359                                                    final List<BinaryDomainCombination> all_bin_domain_combinations_changed,
 360                                                    final int sum_of_all_domains_encountered,
 361                                                    final SortedSet<BinaryDomainCombination> all_bin_domain_combinations_encountered,
 362                                                    final boolean is_gains_analysis ) throws IOException {
 363         SurfacingUtil.checkForOutputFileWriteability( output_file );
 364         final Writer out = ForesterUtil.createBufferedWriter( output_file );
 365         final SortedMap<Object, Integer> bdc_to_counts = ForesterUtil
 366                 .listToSortedCountsMap( all_bin_domain_combinations_changed );
 367         final SortedSet<DomainId> all_domains_in_combination_changed_more_than_once = new TreeSet<DomainId>();
 368         final SortedSet<DomainId> all_domains_in_combination_changed_only_once = new TreeSet<DomainId>();
 369         int above_one = 0;
 370         int one = 0;
 371         for( final Object bdc_object : bdc_to_counts.keySet() ) {
 372             final BinaryDomainCombination bdc = ( BinaryDomainCombination ) bdc_object;
 373             final int count = bdc_to_counts.get( bdc_object );
 374             if ( count < 1 ) {
 375                 ForesterUtil.unexpectedFatalError( PRG_NAME, "count < 1 " );
 376             }
 377             out.write( bdc + "\t" + count + ForesterUtil.LINE_SEPARATOR );
 378             if ( count > 1 ) {
 379                 all_domains_in_combination_changed_more_than_once.add( bdc.getId0() );
 380                 all_domains_in_combination_changed_more_than_once.add( bdc.getId1() );
 381                 above_one++;
 382             }
 383             else if ( count == 1 ) {
 384                 all_domains_in_combination_changed_only_once.add( bdc.getId0() );
 385                 all_domains_in_combination_changed_only_once.add( bdc.getId1() );
 386                 one++;
 387             }
 388         }
 389         final int all = all_bin_domain_combinations_encountered.size();
 390         int never_lost = -1;
 391         if ( !is_gains_analysis ) {
 392             all_bin_domain_combinations_encountered.removeAll( all_bin_domain_combinations_changed );
 393             never_lost = all_bin_domain_combinations_encountered.size();
 394             for( final BinaryDomainCombination bdc : all_bin_domain_combinations_encountered ) {
 395                 out.write( bdc + "\t" + "0" + ForesterUtil.LINE_SEPARATOR );
 396             }
 397         }
 398         if ( is_gains_analysis ) {
 399             out.write( "Sum of all distinct domain combinations appearing once               : " + one
 400                     + ForesterUtil.LINE_SEPARATOR );
 401             out.write( "Sum of all distinct domain combinations appearing more than once     : " + above_one
 402                     + ForesterUtil.LINE_SEPARATOR );
 403             out.write( "Sum of all distinct domains in combinations apppearing only once     : "
 404                     + all_domains_in_combination_changed_only_once.size() + ForesterUtil.LINE_SEPARATOR );
 405             out.write( "Sum of all distinct domains in combinations apppearing more than once: "
 406                     + all_domains_in_combination_changed_more_than_once.size() + ForesterUtil.LINE_SEPARATOR );
 407         }
 408         else {
 409             out.write( "Sum of all distinct domain combinations never lost                   : " + never_lost
 410                     + ForesterUtil.LINE_SEPARATOR );
 411             out.write( "Sum of all distinct domain combinations lost once                    : " + one
 412                     + ForesterUtil.LINE_SEPARATOR );
 413             out.write( "Sum of all distinct domain combinations lost more than once          : " + above_one
 414                     + ForesterUtil.LINE_SEPARATOR );
 415             out.write( "Sum of all distinct domains in combinations lost only once           : "
 416                     + all_domains_in_combination_changed_only_once.size() + ForesterUtil.LINE_SEPARATOR );
 417             out.write( "Sum of all distinct domains in combinations lost more than once: "
 418                     + all_domains_in_combination_changed_more_than_once.size() + ForesterUtil.LINE_SEPARATOR );
 419         }
 420         out.write( "All binary combinations                                              : " + all
 421                 + ForesterUtil.LINE_SEPARATOR );
 422         out.write( "All domains                                                          : "
 423                 + sum_of_all_domains_encountered );
 424         out.close();
 425         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME,
 426                                      "Wrote fitch domain combination dynamics counts analysis to \"" + output_file
 427                                              + "\"" );
 428     }
 429
 430     private static void executePlusMinusAnalysis( final File output_file,
 431                                                   final List<String> plus_minus_analysis_high_copy_base,
 432                                                   final List<String> plus_minus_analysis_high_copy_target,
 433                                                   final List<String> plus_minus_analysis_low_copy,
 434                                                   final List<GenomeWideCombinableDomains> gwcd_list,
 435                                                   final SortedMap<Species, List<Protein>> protein_lists_per_species,
 436                                                   final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 437                                                   final Map<GoId, GoTerm> go_id_to_term_map,
 438                                                   final List<Object> plus_minus_analysis_numbers ) {
 439         final Set<String> all_spec = new HashSet<String>();
 440         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
 441             all_spec.add( gwcd.getSpecies().getSpeciesId() );
 442         }
 443         final File html_out_dom = new File( output_file + PLUS_MINUS_DOM_SUFFIX_HTML );
 444         final File plain_out_dom = new File( output_file + PLUS_MINUS_DOM_SUFFIX );
 445         final File html_out_dc = new File( output_file + PLUS_MINUS_DC_SUFFIX_HTML );
 446         final File all_domains_go_ids_out_dom = new File( output_file + PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX );
 447         final File passing_domains_go_ids_out_dom = new File( output_file + PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX );
 448         final File proteins_file_base = new File( output_file + "" );
 449         final int min_diff = ( ( Integer ) plus_minus_analysis_numbers.get( 0 ) ).intValue();
 450         final double factor = ( ( Double ) plus_minus_analysis_numbers.get( 1 ) ).doubleValue();
 451         try {
 452             DomainCountsDifferenceUtil.calculateCopyNumberDifferences( gwcd_list,
 453                                                                        protein_lists_per_species,
 454                                                                        plus_minus_analysis_high_copy_base,
 455                                                                        plus_minus_analysis_high_copy_target,
 456                                                                        plus_minus_analysis_low_copy,
 457                                                                        min_diff,
 458                                                                        factor,
 459                                                                        plain_out_dom,
 460                                                                        html_out_dom,
 461                                                                        html_out_dc,
 462                                                                        domain_id_to_go_ids_map,
 463                                                                        go_id_to_term_map,
 464                                                                        all_domains_go_ids_out_dom,
 465                                                                        passing_domains_go_ids_out_dom,
 466                                                                        proteins_file_base );
 467         }
 468         catch ( final IOException e ) {
 469             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getLocalizedMessage() );
 470         }
 471         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME, "Wrote plus minus domain analysis results to \""
 472                 + html_out_dom + "\"" );
 473         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME, "Wrote plus minus domain analysis results to \""
 474                 + plain_out_dom + "\"" );
 475         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME, "Wrote plus minus domain analysis results to \""
 476                 + html_out_dc + "\"" );
 477         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME,
 478                                      "Wrote plus minus domain analysis based passing GO ids to \""
 479                                              + passing_domains_go_ids_out_dom + "\"" );
 480         ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME,
 481                                      "Wrote plus minus domain analysis based all GO ids to \""
 482                                              + all_domains_go_ids_out_dom + "\"" );
 483     }
 484
 485     private static Phylogeny[] getIntrees( final File[] intree_files,
 486                                            final int number_of_genomes,
 487                                            final String[][] input_file_properties ) {
 488         final Phylogeny[] intrees = new Phylogeny[ intree_files.length ];
 489         int i = 0;
 490         for( final File intree_file : intree_files ) {
 491             Phylogeny intree = null;
 492             final String error = ForesterUtil.isReadableFile( intree_file );
 493             if ( !ForesterUtil.isEmpty( error ) ) {
 494                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read input tree file [" + intree_file
 495                         + "]: " + error );
 496             }
 497             try {
 498                 final Phylogeny[] p_array = ParserBasedPhylogenyFactory.getInstance()
 499                         .create( intree_file, ParserUtils.createParserDependingOnFileType( intree_file, true ) );
 500                 if ( p_array.length < 1 ) {
 501                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "file [" + intree_file
 502                             + "] does not contain any phylogeny in phyloXML format" );
 503                 }
 504                 else if ( p_array.length > 1 ) {
 505                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "file [" + intree_file
 506                             + "] contains more than one phylogeny in phyloXML format" );
 507                 }
 508                 intree = p_array[ 0 ];
 509             }
 510             catch ( final Exception e ) {
 511                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "failed to read input tree from file ["
 512                         + intree_file + "]: " + error );
 513             }
 514             if ( ( intree == null ) || intree.isEmpty() ) {
 515                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "input tree [" + intree_file + "] is empty" );
 516             }
 517             if ( !intree.isRooted() ) {
 518                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "input tree [" + intree_file + "] is not rooted" );
 519             }
 520             if ( intree.getNumberOfExternalNodes() < number_of_genomes ) {
 521                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 522                                          "number of external nodes [" + intree.getNumberOfExternalNodes()
 523                                                  + "] of input tree [" + intree_file
 524                                                  + "] is smaller than the number of genomes the be analyzed ["
 525                                                  + number_of_genomes + "]" );
 526             }
 527             final StringBuilder parent_names = new StringBuilder();
 528             final int nodes_lacking_name = SurfacingUtil.getNumberOfNodesLackingName( intree, parent_names );
 529             if ( nodes_lacking_name > 0 ) {
 530                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "input tree [" + intree_file + "] has "
 531                         + nodes_lacking_name + " node(s) lacking a name [parent names:" + parent_names + "]" );
 532             }
 533             preparePhylogenyForParsimonyAnalyses( intree, input_file_properties );
 534             if ( !intree.isCompletelyBinary() ) {
 535                 ForesterUtil.printWarningMessage( surfacing_hmmpfam.PRG_NAME, "input tree [" + intree_file
 536                         + "] is not completely binary" );
 537             }
 538             intrees[ i++ ] = intree;
 539         }
 540         return intrees;
 541     }
 542
 543     private static List<Phylogeny> inferSpeciesTrees( final File outfile, final List<DistanceMatrix> distances_list ) {
 544         final NeighborJoining nj = NeighborJoining.createInstance();
 545         final List<Phylogeny> phylogenies = nj.execute( distances_list );
 546         final PhylogenyWriter w = new PhylogenyWriter();
 547         try {
 548             w.toNewHampshire( phylogenies, true, true, outfile, ";" );
 549         }
 550         catch ( final IOException e ) {
 551             ForesterUtil.fatalError( PRG_NAME, "failed to write to outfile [" + outfile + "]: " + e.getMessage() );
 552         }
 553         return phylogenies;
 554     }
 555
 556     public static void main( final String args[] ) {
 557         final long start_time = new Date().getTime();
 558         // final StringBuffer log = new StringBuffer();
 559         final StringBuilder html_desc = new StringBuilder();
 560         ForesterUtil.printProgramInformation( surfacing_hmmpfam.PRG_NAME,
 561                                               surfacing_hmmpfam.PRG_VERSION,
 562                                               surfacing_hmmpfam.PRG_DATE,
 563                                               surfacing_hmmpfam.E_MAIL,
 564                                               surfacing_hmmpfam.WWW );
 565         final String nl = ForesterUtil.LINE_SEPARATOR;
 566         html_desc.append( "<table>" + nl );
 567         html_desc.append( "<tr><td>Produced by:</td><td>" + surfacing_hmmpfam.PRG_NAME + "</td></tr>" + nl );
 568         html_desc.append( "<tr><td>Version:</td><td>" + surfacing_hmmpfam.PRG_VERSION + "</td></tr>" + nl );
 569         html_desc.append( "<tr><td>Release Date:</td><td>" + surfacing_hmmpfam.PRG_DATE + "</td></tr>" + nl );
 570         html_desc.append( "<tr><td>Contact:</td><td>" + surfacing_hmmpfam.E_MAIL + "</td></tr>" + nl );
 571         html_desc.append( "<tr><td>WWW:</td><td>" + surfacing_hmmpfam.WWW + "</td></tr>" + nl );
 572         CommandLineArguments cla = null;
 573         try {
 574             cla = new CommandLineArguments( args );
 575         }
 576         catch ( final Exception e ) {
 577             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage() );
 578         }
 579         if ( cla.isOptionSet( surfacing_hmmpfam.HELP_OPTION_1 ) || cla.isOptionSet( surfacing_hmmpfam.HELP_OPTION_2 ) ) {
 580             surfacing_hmmpfam.printHelp();
 581             System.exit( 0 );
 582         }
 583         if ( ( args.length < 1 ) ) {
 584             surfacing_hmmpfam.printHelp();
 585             System.exit( -1 );
 586         }
 587         final List<String> allowed_options = new ArrayList<String>();
 588         allowed_options.add( surfacing_hmmpfam.NOT_IGNORE_DUFS_OPTION );
 589         allowed_options.add( surfacing_hmmpfam.MAX_E_VALUE_OPTION );
 590         allowed_options.add( surfacing_hmmpfam.DETAILEDNESS_OPTION );
 591         allowed_options.add( surfacing_hmmpfam.OUTPUT_FILE_OPTION );
 592         allowed_options.add( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION );
 593         allowed_options.add( surfacing_hmmpfam.SPECIES_MATRIX_OPTION );
 594         allowed_options.add( surfacing_hmmpfam.SCORING_OPTION );
 595         allowed_options.add( surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_OPTION );
 596         allowed_options.add( surfacing_hmmpfam.NO_ENGULFING_OVERLAP_OPTION );
 597         allowed_options.add( surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION );
 598         allowed_options.add( surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION );
 599         allowed_options.add( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION );
 600         allowed_options.add( surfacing_hmmpfam.OUTPUT_DIR_OPTION );
 601         allowed_options.add( surfacing_hmmpfam.IGNORE_COMBINATION_WITH_SAME_OPTION );
 602         allowed_options.add( surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION );
 603         allowed_options.add( surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION );
 604         allowed_options.add( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION );
 605         allowed_options.add( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION );
 606         allowed_options.add( surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 607         allowed_options.add( surfacing_hmmpfam.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION );
 608         allowed_options.add( surfacing_hmmpfam.DISPLAY_M_HISTOGRAMS_OPTION );
 609         allowed_options.add( surfacing_hmmpfam.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS );
 610         allowed_options.add( JACKNIFE_OPTION );
 611         allowed_options.add( JACKNIFE_RANDOM_SEED_OPTION );
 612         allowed_options.add( JACKNIFE_RATIO_OPTION );
 613         allowed_options.add( INPUT_SPECIES_TREE_OPTION );
 614         //allowed_options.add( INFER_SPECIES_TREES_OPTION );
 615         allowed_options.add( FILTER_POSITIVE_OPTION );
 616         allowed_options.add( FILTER_NEGATIVE_OPTION );
 617         allowed_options.add( INPUT_FILES_FROM_FILE_OPTION );
 618         allowed_options.add( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
 619         allowed_options.add( FILTER_NEGATIVE_DOMAINS_OPTION );
 620         allowed_options.add( IGNORE_VIRAL_IDS );
 621         allowed_options.add( SEQ_EXTRACT_OPTION );
 622         allowed_options.add( SECONDARY_FEATURES_PARSIMONY_MAP_FILE );
 623         allowed_options.add( PLUS_MINUS_ANALYSIS_OPTION );
 624         allowed_options.add( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS );
 625         allowed_options.add( OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS );
 626         allowed_options.add( CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY );
 627         boolean ignore_dufs = surfacing_hmmpfam.IGNORE_DUFS_DEFAULT;
 628         boolean ignore_combination_with_same = surfacing_hmmpfam.IGNORE_COMBINATION_WITH_SAME_DEFAULLT;
 629         double e_value_max = surfacing_hmmpfam.MAX_E_VALUE_DEFAULT;
 630         int max_allowed_overlap = surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_DEFAULT;
 631         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 632         if ( dissallowed_options.length() > 0 ) {
 633             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown option(s): " + dissallowed_options );
 634         }
 635         boolean output_binary_domain_combinationsfor_graph_analysis = false;
 636         if ( cla.isOptionSet( DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS ) ) {
 637             output_binary_domain_combinationsfor_graph_analysis = true;
 638         }
 639         if ( cla.isOptionSet( surfacing_hmmpfam.MAX_E_VALUE_OPTION ) ) {
 640             try {
 641                 e_value_max = cla.getOptionValueAsDouble( surfacing_hmmpfam.MAX_E_VALUE_OPTION );
 642             }
 643             catch ( final Exception e ) {
 644                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no acceptable value for E-value maximum" );
 645             }
 646         }
 647         if ( cla.isOptionSet( surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_OPTION ) ) {
 648             try {
 649                 max_allowed_overlap = cla.getOptionValueAsInt( surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_OPTION );
 650             }
 651             catch ( final Exception e ) {
 652                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 653                                          "no acceptable value for maximal allowed domain overlap" );
 654             }
 655         }
 656         boolean no_engulfing_overlaps = false;
 657         if ( cla.isOptionSet( surfacing_hmmpfam.NO_ENGULFING_OVERLAP_OPTION ) ) {
 658             no_engulfing_overlaps = true;
 659         }
 660         boolean ignore_virus_like_ids = false;
 661         if ( cla.isOptionSet( surfacing_hmmpfam.IGNORE_VIRAL_IDS ) ) {
 662             ignore_virus_like_ids = true;
 663         }
 664         if ( cla.isOptionSet( surfacing_hmmpfam.NOT_IGNORE_DUFS_OPTION ) ) {
 665             ignore_dufs = false;
 666         }
 667         if ( cla.isOptionSet( surfacing_hmmpfam.IGNORE_COMBINATION_WITH_SAME_OPTION ) ) {
 668             ignore_combination_with_same = true;
 669         }
 670         boolean ignore_domains_without_combs_in_all_spec = IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT;
 671         if ( cla.isOptionSet( surfacing_hmmpfam.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION ) ) {
 672             ignore_domains_without_combs_in_all_spec = true;
 673         }
 674         boolean ignore_species_specific_domains = IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT;
 675         if ( cla.isOptionSet( surfacing_hmmpfam.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION ) ) {
 676             ignore_species_specific_domains = true;
 677         }
 678         File output_file = null;
 679         if ( cla.isOptionSet( surfacing_hmmpfam.OUTPUT_FILE_OPTION ) ) {
 680             if ( !cla.isOptionValueSet( surfacing_hmmpfam.OUTPUT_FILE_OPTION ) ) {
 681                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 682                                          "no value for domain combinations similarities output file: -"
 683                                                  + surfacing_hmmpfam.OUTPUT_FILE_OPTION + "=<file>" );
 684             }
 685             output_file = new File( cla.getOptionValue( surfacing_hmmpfam.OUTPUT_FILE_OPTION ) );
 686             SurfacingUtil.checkForOutputFileWriteability( output_file );
 687         }
 688         File cutoff_scores_file = null;
 689         Map<String, String> individual_domain_score_cutoffs = null;
 690         if ( cla.isOptionSet( surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION ) ) {
 691             if ( !cla.isOptionValueSet( surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION ) ) {
 692                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 693                                          "no value for individual domain score cutoffs file: -"
 694                                                  + surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION + "=<file>" );
 695             }
 696             cutoff_scores_file = new File( cla.getOptionValue( surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION ) );
 697             final String error = ForesterUtil.isReadableFile( cutoff_scores_file );
 698             if ( !ForesterUtil.isEmpty( error ) ) {
 699                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 700                                          "cannot read individual domain score cutoffs file: " + error );
 701             }
 702             try {
 703                 final BasicTable<String> scores_table = BasicTableParser.parse( cutoff_scores_file, " " );
 704                 individual_domain_score_cutoffs = scores_table.getColumnsAsMap( 0, 1 );
 705             }
 706             catch ( final IOException e ) {
 707                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 708                                          "cannot read from individual domain score cutoffs file: " + e );
 709             }
 710         }
 711         BinaryDomainCombination.DomainCombinationType dc_type = BinaryDomainCombination.DomainCombinationType.BASIC;
 712         if ( cla.isOptionSet( surfacing_hmmpfam.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS ) ) {
 713             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED;
 714         }
 715         if ( cla.isOptionSet( surfacing_hmmpfam.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY ) ) {
 716             dc_type = BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT;
 717         }
 718         File out_dir = null;
 719         if ( cla.isOptionSet( surfacing_hmmpfam.OUTPUT_DIR_OPTION ) ) {
 720             if ( !cla.isOptionValueSet( surfacing_hmmpfam.OUTPUT_DIR_OPTION ) ) {
 721                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for output directory: -"
 722                         + surfacing_hmmpfam.OUTPUT_DIR_OPTION + "=<dir>" );
 723             }
 724             out_dir = new File( cla.getOptionValue( surfacing_hmmpfam.OUTPUT_DIR_OPTION ) );
 725             if ( out_dir.exists() && ( out_dir.listFiles().length > 0 ) ) {
 726                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "\"" + out_dir
 727                         + "\" aready exists and is not empty" );
 728             }
 729             if ( !out_dir.exists() ) {
 730                 final boolean success = out_dir.mkdir();
 731                 if ( !success || !out_dir.exists() ) {
 732                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "failed to create \"" + out_dir + "\"" );
 733                 }
 734             }
 735             if ( !out_dir.canWrite() ) {
 736                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot write to \"" + out_dir + "\"" );
 737             }
 738         }
 739         File positive_filter_file = null;
 740         File negative_filter_file = null;
 741         File negative_domains_filter_file = null;
 742         if ( cla.isOptionSet( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION )
 743                 && cla.isOptionSet( surfacing_hmmpfam.FILTER_POSITIVE_OPTION ) ) {
 744             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 745                                      "attempt to use both negative and positive protein filter" );
 746         }
 747         if ( cla.isOptionSet( surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION )
 748                 && ( cla.isOptionSet( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION ) || cla
 749                         .isOptionSet( surfacing_hmmpfam.FILTER_POSITIVE_OPTION ) ) ) {
 750             ForesterUtil
 751                     .fatalError( surfacing_hmmpfam.PRG_NAME,
 752                                  "attempt to use both negative or positive protein filter together wirh a negative domains filter" );
 753         }
 754         if ( cla.isOptionSet( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION ) ) {
 755             if ( !cla.isOptionValueSet( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION ) ) {
 756                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for negative filter: -"
 757                         + surfacing_hmmpfam.FILTER_NEGATIVE_OPTION + "=<file>" );
 758             }
 759             negative_filter_file = new File( cla.getOptionValue( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION ) );
 760             final String msg = ForesterUtil.isReadableFile( negative_filter_file );
 761             if ( !ForesterUtil.isEmpty( msg ) ) {
 762                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "can not read from \"" + negative_filter_file
 763                         + "\": " + msg );
 764             }
 765         }
 766         else if ( cla.isOptionSet( surfacing_hmmpfam.FILTER_POSITIVE_OPTION ) ) {
 767             if ( !cla.isOptionValueSet( surfacing_hmmpfam.FILTER_POSITIVE_OPTION ) ) {
 768                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for positive filter: -"
 769                         + surfacing_hmmpfam.FILTER_POSITIVE_OPTION + "=<file>" );
 770             }
 771             positive_filter_file = new File( cla.getOptionValue( surfacing_hmmpfam.FILTER_POSITIVE_OPTION ) );
 772             final String msg = ForesterUtil.isReadableFile( positive_filter_file );
 773             if ( !ForesterUtil.isEmpty( msg ) ) {
 774                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "can not read from \"" + positive_filter_file
 775                         + "\": " + msg );
 776             }
 777         }
 778         else if ( cla.isOptionSet( surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 779             if ( !cla.isOptionValueSet( surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION ) ) {
 780                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for negative domains filter: -"
 781                         + surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION + "=<file>" );
 782             }
 783             negative_domains_filter_file = new File( cla.getOptionValue( surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION ) );
 784             final String msg = ForesterUtil.isReadableFile( negative_domains_filter_file );
 785             if ( !ForesterUtil.isEmpty( msg ) ) {
 786                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "can not read from \""
 787                         + negative_domains_filter_file + "\": " + msg );
 788             }
 789         }
 790         final List<String> plus_minus_analysis_high_copy_base_species = new ArrayList<String>();
 791         final List<String> plus_minus_analysis_high_copy_target_species = new ArrayList<String>();
 792         final List<String> plus_minus_analysis_high_low_copy_species = new ArrayList<String>();
 793         final List<Object> plus_minus_analysis_numbers = new ArrayList<Object>();
 794         processPlusMinusAnalysisOption( cla,
 795                                         plus_minus_analysis_high_copy_base_species,
 796                                         plus_minus_analysis_high_copy_target_species,
 797                                         plus_minus_analysis_high_low_copy_species,
 798                                         plus_minus_analysis_numbers );
 799         File input_files_file = null;
 800         String[] input_file_names_from_file = null;
 801         if ( cla.isOptionSet( surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION ) ) {
 802             if ( !cla.isOptionValueSet( surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION ) ) {
 803                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for input files file: -"
 804                         + surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION + "=<file>" );
 805             }
 806             input_files_file = new File( cla.getOptionValue( surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION ) );
 807             final String msg = ForesterUtil.isReadableFile( input_files_file );
 808             if ( !ForesterUtil.isEmpty( msg ) ) {
 809                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "can not read from \"" + input_files_file + "\": "
 810                         + msg );
 811             }
 812             try {
 813                 input_file_names_from_file = ForesterUtil.file2array( input_files_file );
 814             }
 815             catch ( final IOException e ) {
 816                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "failed to read from \"" + input_files_file
 817                         + "\": " + e );
 818             }
 819         }
 820         if ( ( cla.getNumberOfNames() < 1 )
 821                 && ( ( input_file_names_from_file == null ) || ( input_file_names_from_file.length < 1 ) ) ) {
 822             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 823                                      "No hmmpfam output file indicated is input: use comand line directly or "
 824                                              + surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION + "=<file>" );
 825         }
 826         DomainSimilarity.DomainSimilarityScoring scoring = SCORING_DEFAULT;
 827         if ( cla.isOptionSet( surfacing_hmmpfam.SCORING_OPTION ) ) {
 828             if ( !cla.isOptionValueSet( surfacing_hmmpfam.SCORING_OPTION ) ) {
 829                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 830                                          "no value for scoring method for domain combinations similarity calculation: -"
 831                                                  + surfacing_hmmpfam.SCORING_OPTION + "=<"
 832                                                  + surfacing_hmmpfam.SCORING_DOMAIN_COUNT_BASED + "|"
 833                                                  + surfacing_hmmpfam.SCORING_PROTEIN_COUNT_BASED + "|"
 834                                                  + surfacing_hmmpfam.SCORING_COMBINATION_BASED + ">\"" );
 835             }
 836             final String scoring_str = cla.getOptionValue( surfacing_hmmpfam.SCORING_OPTION );
 837             if ( scoring_str.equals( surfacing_hmmpfam.SCORING_DOMAIN_COUNT_BASED ) ) {
 838                 scoring = DomainSimilarity.DomainSimilarityScoring.DOMAINS;
 839             }
 840             else if ( scoring_str.equals( surfacing_hmmpfam.SCORING_COMBINATION_BASED ) ) {
 841                 scoring = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
 842             }
 843             else if ( scoring_str.equals( surfacing_hmmpfam.SCORING_PROTEIN_COUNT_BASED ) ) {
 844                 scoring = DomainSimilarity.DomainSimilarityScoring.PROTEINS;
 845             }
 846             else {
 847                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + scoring_str
 848                         + "\" for scoring method for domain combinations similarity calculation: \"-"
 849                         + surfacing_hmmpfam.SCORING_OPTION + "=<" + surfacing_hmmpfam.SCORING_DOMAIN_COUNT_BASED + "|"
 850                         + surfacing_hmmpfam.SCORING_PROTEIN_COUNT_BASED + "|"
 851                         + surfacing_hmmpfam.SCORING_COMBINATION_BASED + ">\"" );
 852             }
 853         }
 854         boolean sort_by_species_count_first = false;
 855         if ( cla.isOptionSet( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION ) ) {
 856             sort_by_species_count_first = true;
 857         }
 858         boolean species_matrix = false;
 859         if ( cla.isOptionSet( surfacing_hmmpfam.SPECIES_MATRIX_OPTION ) ) {
 860             species_matrix = true;
 861         }
 862         boolean output_protein_lists_for_all_domains = false;
 863         if ( cla.isOptionSet( surfacing_hmmpfam.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS ) ) {
 864             output_protein_lists_for_all_domains = true;
 865         }
 866         Detailedness detailedness = DETAILEDNESS_DEFAULT;
 867         if ( cla.isOptionSet( surfacing_hmmpfam.DETAILEDNESS_OPTION ) ) {
 868             if ( !cla.isOptionValueSet( surfacing_hmmpfam.DETAILEDNESS_OPTION ) ) {
 869                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for -"
 870                         + surfacing_hmmpfam.DETAILEDNESS_OPTION + "=<" + surfacing_hmmpfam.DETAILEDNESS_BASIC + "|"
 871                         + surfacing_hmmpfam.DETAILEDNESS_LIST_IDS + "|" + surfacing_hmmpfam.DETAILEDNESS_PUNCTILIOUS
 872                         + ">\"" );
 873             }
 874             final String detness = cla.getOptionValue( surfacing_hmmpfam.DETAILEDNESS_OPTION ).toLowerCase();
 875             if ( detness.equals( surfacing_hmmpfam.DETAILEDNESS_BASIC ) ) {
 876                 detailedness = DomainSimilarityCalculator.Detailedness.BASIC;
 877             }
 878             else if ( detness.equals( surfacing_hmmpfam.DETAILEDNESS_LIST_IDS ) ) {
 879                 detailedness = DomainSimilarityCalculator.Detailedness.LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES;
 880             }
 881             else if ( detness.equals( surfacing_hmmpfam.DETAILEDNESS_PUNCTILIOUS ) ) {
 882                 detailedness = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
 883             }
 884             else {
 885                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + detness
 886                         + "\" for detailedness: \"-" + surfacing_hmmpfam.DETAILEDNESS_OPTION + "=<"
 887                         + surfacing_hmmpfam.DETAILEDNESS_BASIC + "|" + surfacing_hmmpfam.DETAILEDNESS_LIST_IDS + "|"
 888                         + surfacing_hmmpfam.DETAILEDNESS_PUNCTILIOUS + ">\"" );
 889             }
 890         }
 891         String automated_pairwise_comparison_suffix = null;
 892         boolean perform_pwc = false;
 893         boolean write_pwc_files = false;
 894         if ( cla.isOptionSet( surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 895             perform_pwc = true;
 896             if ( !cla.isOptionValueSet( surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION ) ) {
 897                 write_pwc_files = false;
 898             }
 899             else {
 900                 write_pwc_files = true;
 901                 automated_pairwise_comparison_suffix = "_"
 902                         + cla.getOptionValue( surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION );
 903             }
 904         }
 905         String query_domain_ids = null;
 906         if ( cla.isOptionSet( surfacing_hmmpfam.SEQ_EXTRACT_OPTION ) ) {
 907             if ( !cla.isOptionValueSet( surfacing_hmmpfam.SEQ_EXTRACT_OPTION ) ) {
 908                 ForesterUtil
 909                         .fatalError( surfacing_hmmpfam.PRG_NAME,
 910                                      "no domain ids given for sequences with given domains to be extracted : -"
 911                                              + surfacing_hmmpfam.SEQ_EXTRACT_OPTION
 912                                              + "=<ordered domain sequences, domain ids separated by '~', sequences separated by '#'>" );
 913             }
 914             query_domain_ids = cla.getOptionValue( surfacing_hmmpfam.SEQ_EXTRACT_OPTION );
 915         }
 916         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field = DOMAIN_SORT_FILD_DEFAULT;
 917         DomainSimilarity.DomainSimilaritySortField domain_similarity_sort_field_for_automated_pwc = DOMAIN_SORT_FILD_DEFAULT;
 918         if ( cla.isOptionSet( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 919             if ( !cla.isOptionValueSet( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION ) ) {
 920                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
 921                                          "no value for domain combinations similarities sorting: -"
 922                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION + "=<"
 923                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 924                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MAX + "|"
 925                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MIN + "|"
 926                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MEAN + "|"
 927                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 928                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 929                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|"
 930                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|"
 931                                                  + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SD + ">\"" );
 932             }
 933             final String sort_str = cla.getOptionValue( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION ).toLowerCase();
 934             if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ALPHA ) ) {
 935                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 936                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 937             }
 938             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MAX ) ) {
 939                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX;
 940                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 941             }
 942             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MIN ) ) {
 943                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MIN;
 944                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 945             }
 946             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MEAN ) ) {
 947                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MEAN;
 948                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MEAN;
 949             }
 950             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT ) ) {
 951                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SPECIES_COUNT;
 952                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 953             }
 954             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SD ) ) {
 955                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.SD;
 956                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
 957             }
 958             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_DIFF ) ) {
 959                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 960                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_DIFFERENCE;
 961             }
 962             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF ) ) {
 963                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 964                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
 965             }
 966             else if ( sort_str.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF ) ) {
 967                 domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 968                 domain_similarity_sort_field_for_automated_pwc = DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE;
 969             }
 970             else {
 971                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + sort_str
 972                         + "\" for domain combinations similarities sorting: \"-"
 973                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION + "=<"
 974                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ALPHA + "|"
 975                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MAX + "|"
 976                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MIN + "|"
 977                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_MEAN + "|"
 978                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_DIFF + "|"
 979                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF + "|"
 980                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_COUNTS_DIFF + "|" + "|"
 981                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SPECIES_COUNT + "|"
 982                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_SD + ">\"" );
 983             }
 984         }
 985         PrintableDomainSimilarity.PRINT_OPTION domain_similarity_print_option = DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT;
 986         if ( cla.isOptionSet( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 987             if ( !cla.isOptionValueSet( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION ) ) {
 988                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for print option: -"
 989                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML + "|"
 990                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
 991                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
 992             }
 993             final String sort = cla.getOptionValue( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION ).toLowerCase();
 994             if ( sort.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML ) ) {
 995                 domain_similarity_print_option = PrintableDomainSimilarity.PRINT_OPTION.HTML;
 996             }
 997             else if ( sort.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML ) ) {
 998                 // domain_similarity_print_option =
 999                 // DomainSimilarity.PRINT_OPTION.SIMPLE_HTML;
1000                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "simple HTML output not implemented yet :(" );
1001             }
1002             else if ( sort.equals( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED ) ) {
1003                 domain_similarity_print_option = PrintableDomainSimilarity.PRINT_OPTION.SIMPLE_TAB_DELIMITED;
1004             }
1005             else {
1006                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + sort
1007                         + "\" for print option: -" + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML
1008                         + "|" + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML + "|"
1009                         + surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED + ">\"" );
1010             }
1011         }
1012         GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order = DOMAINS_SORT_ORDER_DEFAULT;
1013         if ( cla.isOptionSet( surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION ) ) {
1014             if ( !cla.isOptionValueSet( surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION ) ) {
1015                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for sorting of domain counts: -"
1016                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION + "=<" + surfacing_hmmpfam.DOMAIN_COUNT_SORT_ALPHA
1017                         + "|" + surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
1018                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
1019                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
1020             }
1021             final String sort = cla.getOptionValue( surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION ).toLowerCase();
1022             if ( sort.equals( surfacing_hmmpfam.DOMAIN_COUNT_SORT_ALPHA ) ) {
1023                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
1024             }
1025             else if ( sort.equals( surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT ) ) {
1026                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_COUNT;
1027             }
1028             else if ( sort.equals( surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT ) ) {
1029                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.KEY_DOMAIN_PROTEINS_COUNT;
1030             }
1031             else if ( sort.equals( surfacing_hmmpfam.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT ) ) {
1032                 dc_sort_order = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.COMBINATIONS_COUNT;
1033             }
1034             else {
1035                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + sort
1036                         + "\" for sorting of domain counts: \"-" + surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION + "=<"
1037                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_ALPHA + "|"
1038                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT + "|"
1039                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT + "|"
1040                         + surfacing_hmmpfam.DOMAIN_COUNT_SORT_COMBINATIONS_COUNT + ">\"" );
1041             }
1042         }
1043         String[][] input_file_properties = null;
1044         if ( input_file_names_from_file != null ) {
1045             input_file_properties = surfacing_hmmpfam.processInputFileNames( input_file_names_from_file );
1046         }
1047         else {
1048             input_file_properties = surfacing_hmmpfam.processInputFileNames( cla.getNames() );
1049         }
1050         final int number_of_genomes = input_file_properties.length;
1051         if ( number_of_genomes < 2 ) {
1052             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot analyze less than two files" );
1053         }
1054         if ( ( number_of_genomes < 3 ) && perform_pwc ) {
1055             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot use : -"
1056                     + surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1057                     + "=<suffix> to turn on pairwise analyses with less than three input files" );
1058         }
1059         checkWriteabilityForPairwiseComparisons( domain_similarity_print_option,
1060                                                  input_file_properties,
1061                                                  automated_pairwise_comparison_suffix,
1062                                                  out_dir );
1063         for( int i = 0; i < number_of_genomes; i++ ) {
1064             File dcc_outfile = new File( input_file_properties[ i ][ 0 ]
1065                     + surfacing_hmmpfam.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
1066             if ( out_dir != null ) {
1067                 dcc_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + dcc_outfile );
1068             }
1069             SurfacingUtil.checkForOutputFileWriteability( dcc_outfile );
1070         }
1071         File pfam_to_go_file = null;
1072         Map<DomainId, List<GoId>> domain_id_to_go_ids_map = null;
1073         int domain_id_to_go_ids_count = 0;
1074         if ( cla.isOptionSet( surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION ) ) {
1075             if ( !cla.isOptionValueSet( surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION ) ) {
1076                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for Pfam to GO mapping file: -"
1077                         + surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION + "=<file>" );
1078             }
1079             pfam_to_go_file = new File( cla.getOptionValue( surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION ) );
1080             final String error = ForesterUtil.isReadableFile( pfam_to_go_file );
1081             if ( !ForesterUtil.isEmpty( error ) ) {
1082                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read Pfam to GO mapping file: " + error );
1083             }
1084             try {
1085                 final PfamToGoParser parser = new PfamToGoParser( pfam_to_go_file );
1086                 final List<PfamToGoMapping> pfam_to_go_mappings = parser.parse();
1087                 domain_id_to_go_ids_map = SurfacingUtil.createDomainIdToGoIdMap( pfam_to_go_mappings );
1088                 if ( parser.getMappingCount() < domain_id_to_go_ids_map.size() ) {
1089                     ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME,
1090                                                        "parser.getMappingCount() < domain_id_to_go_ids_map.size()" );
1091                 }
1092                 domain_id_to_go_ids_count = parser.getMappingCount();
1093             }
1094             catch ( final IOException e ) {
1095                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read from Pfam to GO mapping file: " + e );
1096             }
1097         }
1098         File go_obo_file = null;
1099         List<GoTerm> go_terms = null;
1100         if ( cla.isOptionSet( surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION ) ) {
1101             if ( !cla.isOptionValueSet( surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION ) ) {
1102                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for GO OBO file: -"
1103                         + surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION + "=<file>" );
1104             }
1105             if ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) {
1106                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot use GO OBO file (-"
1107                         + surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION + "=<file>) without Pfam to GO mapping file ("
1108                         + surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION + "=<file>)" );
1109             }
1110             go_obo_file = new File( cla.getOptionValue( surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION ) );
1111             final String error = ForesterUtil.isReadableFile( go_obo_file );
1112             if ( !ForesterUtil.isEmpty( error ) ) {
1113                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read GO OBO file: " + error );
1114             }
1115             try {
1116                 final OBOparser parser = new OBOparser( go_obo_file, OBOparser.ReturnType.BASIC_GO_TERM );
1117                 go_terms = parser.parse();
1118                 if ( parser.getGoTermCount() != go_terms.size() ) {
1119                     ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME,
1120                                                        "parser.getGoTermCount() != go_terms.size()" );
1121                 }
1122             }
1123             catch ( final IOException e ) {
1124                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
1125                                          "cannot read from GO OBO file: " + e.getLocalizedMessage() );
1126             }
1127         }
1128         Map<GoId, GoTerm> go_id_to_term_map = null;
1129         if ( ( ( domain_id_to_go_ids_map != null ) && ( domain_id_to_go_ids_map.size() > 0 ) )
1130                 && ( ( go_terms != null ) && ( go_terms.size() > 0 ) ) ) {
1131             go_id_to_term_map = GoUtils.createGoIdToGoTermMap( go_terms );
1132         }
1133         GoNameSpace go_namespace_limit = null;
1134         if ( cla.isOptionSet( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION ) ) {
1135             if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
1136                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot use GO namespace limit (-"
1137                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION
1138                         + "=<namespace>) without Pfam to GO mapping file ("
1139                         + surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION + "=<file>) and GO OBO file (-"
1140                         + surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION + "=<file>)" );
1141             }
1142             if ( !cla.isOptionValueSet( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION ) ) {
1143                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for GO namespace limit: \"-"
1144                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION + "=<"
1145                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
1146                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
1147                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
1148             }
1149             final String go_namespace_limit_str = cla.getOptionValue( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION )
1150                     .toLowerCase();
1151             if ( go_namespace_limit_str.equals( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION ) ) {
1152                 go_namespace_limit = GoNameSpace.createMolecularFunction();
1153             }
1154             else if ( go_namespace_limit_str.equals( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS ) ) {
1155                 go_namespace_limit = GoNameSpace.createBiologicalProcess();
1156             }
1157             else if ( go_namespace_limit_str.equals( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT ) ) {
1158                 go_namespace_limit = GoNameSpace.createCellularComponent();
1159             }
1160             else {
1161                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "unknown value \"" + go_namespace_limit_str
1162                         + "\" for GO namespace limit: \"-" + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION + "=<"
1163                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION + "|"
1164                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS + "|"
1165                         + surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT + ">\"" );
1166             }
1167         }
1168         if ( ( domain_similarity_sort_field == DomainSimilarity.DomainSimilaritySortField.MAX_COUNTS_DIFFERENCE )
1169                 && ( number_of_genomes > 2 ) ) {
1170             domain_similarity_sort_field = DomainSimilarity.DomainSimilaritySortField.ABS_MAX_COUNTS_DIFFERENCE;
1171         }
1172         boolean jacknifed_distances = false;
1173         int jacknife_resamplings = JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT;
1174         double jacknife_ratio = JACKNIFE_RATIO_DEFAULT;
1175         long random_seed = JACKNIFE_RANDOM_SEED_DEFAULT;
1176         if ( cla.isOptionSet( surfacing_hmmpfam.JACKNIFE_OPTION ) ) {
1177             if ( ( number_of_genomes < 3 ) || !perform_pwc ) {
1178                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot use jacknife resampling analysis (-"
1179                         + surfacing_hmmpfam.JACKNIFE_OPTION + "[=<number of resamplings>]) without pairwise analyses ("
1180                         + surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1181                         + "=<suffix for pairwise comparison output files>)" );
1182             }
1183             jacknifed_distances = true;
1184             if ( cla.isOptionHasAValue( surfacing_hmmpfam.JACKNIFE_OPTION ) ) {
1185                 try {
1186                     jacknife_resamplings = cla.getOptionValueAsInt( surfacing_hmmpfam.JACKNIFE_OPTION );
1187                 }
1188                 catch ( final IOException e ) {
1189                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "illegal format for number of resamplings" );
1190                 }
1191                 if ( jacknife_resamplings < 2 ) {
1192                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "attempt to use less than 2 resamplings" );
1193                 }
1194             }
1195             if ( cla.isOptionSet( surfacing_hmmpfam.JACKNIFE_RATIO_OPTION )
1196                     && cla.isOptionHasAValue( surfacing_hmmpfam.JACKNIFE_RATIO_OPTION ) ) {
1197                 try {
1198                     jacknife_ratio = cla.getOptionValueAsDouble( surfacing_hmmpfam.JACKNIFE_RATIO_OPTION );
1199                 }
1200                 catch ( final IOException e ) {
1201                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "illegal format for jacknife ratio" );
1202                 }
1203                 if ( ( jacknife_ratio <= 0.0 ) || ( jacknife_ratio >= 1.0 ) ) {
1204                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
1205                                              "attempt to use illegal value for jacknife ratio: " + jacknife_ratio );
1206                 }
1207             }
1208             if ( cla.isOptionSet( surfacing_hmmpfam.JACKNIFE_RANDOM_SEED_OPTION )
1209                     && cla.isOptionHasAValue( surfacing_hmmpfam.JACKNIFE_RANDOM_SEED_OPTION ) ) {
1210                 try {
1211                     random_seed = cla.getOptionValueAsLong( surfacing_hmmpfam.JACKNIFE_RANDOM_SEED_OPTION );
1212                 }
1213                 catch ( final IOException e ) {
1214                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "illegal format for random generator seed" );
1215                 }
1216             }
1217         }
1218         //        boolean infer_species_trees = false;
1219         //        if ( cla.isOptionSet( surfacing.INFER_SPECIES_TREES_OPTION ) ) {
1220         //            if ( ( output_file == null ) || ( number_of_genomes < 3 )
1221         //                    || ForesterUtil.isEmpty( automated_pairwise_comparison_suffix ) ) {
1222         //                ForesterUtil.fatalError( surfacing.PRG_NAME, "cannot infer species trees (-"
1223         //                        + surfacing.INFER_SPECIES_TREES_OPTION + " without pairwise analyses ("
1224         //                        + surfacing.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1225         //                        + "=<suffix for pairwise comparison output files>)" );
1226         //            }
1227         //            infer_species_trees = true;
1228         //        }
1229         File[] intree_files = null;
1230         Phylogeny[] intrees = null;
1231         if ( cla.isOptionSet( surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION ) ) {
1232             // TODO FIXME if jacknife.... maybe not
1233             if ( number_of_genomes < 3 ) {
1234                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
1235                                          "cannot infer gains and losses on input species trees (-"
1236                                                  + surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION
1237                                                  + " without pairwise analyses ("
1238                                                  + surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION
1239                                                  + "=<suffix for pairwise comparison output files>)" );
1240             }
1241             if ( !cla.isOptionValueSet( surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION ) ) {
1242                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for input tree: -"
1243                         + surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION + "=<tree file in phyloXML format>" );
1244             }
1245             final String intrees_str = cla.getOptionValue( surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION );
1246             if ( intrees_str.indexOf( "#" ) > 0 ) {
1247                 final String[] intrees_strs = intrees_str.split( "#" );
1248                 intree_files = new File[ intrees_strs.length ];
1249                 int i = 0;
1250                 for( final String s : intrees_strs ) {
1251                     intree_files[ i++ ] = new File( s.trim() );
1252                 }
1253             }
1254             else {
1255                 intree_files = new File[ 1 ];
1256                 intree_files[ 0 ] = new File( intrees_str );
1257             }
1258             intrees = getIntrees( intree_files, number_of_genomes, input_file_properties );
1259         }
1260         long random_number_seed_for_fitch_parsimony = 0l;
1261         boolean radomize_fitch_parsimony = false;
1262         if ( cla.isOptionSet( surfacing_hmmpfam.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
1263             if ( !cla.isOptionValueSet( surfacing_hmmpfam.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION ) ) {
1264                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for random number seed: -"
1265                         + surfacing_hmmpfam.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION + "=<seed>" );
1266             }
1267             try {
1268                 random_number_seed_for_fitch_parsimony = cla
1269                         .getOptionValueAsLong( RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION );
1270             }
1271             catch ( final IOException e ) {
1272                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage() );
1273             }
1274             radomize_fitch_parsimony = true;
1275         }
1276         SortedSet<DomainId> filter = null;
1277         if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
1278                 || ( negative_domains_filter_file != null ) ) {
1279             filter = new TreeSet<DomainId>();
1280             if ( positive_filter_file != null ) {
1281                 processFilter( positive_filter_file, filter );
1282             }
1283             else if ( negative_filter_file != null ) {
1284                 processFilter( negative_filter_file, filter );
1285             }
1286             else if ( negative_domains_filter_file != null ) {
1287                 processFilter( negative_domains_filter_file, filter );
1288             }
1289         }
1290         Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps = null;
1291         File[] secondary_features_map_files = null;
1292         final File domain_lengths_analysis_outfile = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
1293                 + DOMAIN_LENGTHS_ANALYSIS_SUFFIX );
1294         if ( PERFORM_DOMAIN_LENGTH_ANALYSIS ) {
1295             SurfacingUtil.checkForOutputFileWriteability( domain_lengths_analysis_outfile );
1296         }
1297         if ( cla.isOptionSet( surfacing_hmmpfam.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
1298             if ( !cla.isOptionValueSet( surfacing_hmmpfam.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ) ) {
1299                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for secondary features map file: -"
1300                         + surfacing_hmmpfam.SECONDARY_FEATURES_PARSIMONY_MAP_FILE + "=<file>" );
1301             }
1302             final String[] secondary_features_map_files_strs = cla
1303                     .getOptionValue( surfacing_hmmpfam.SECONDARY_FEATURES_PARSIMONY_MAP_FILE ).split( "#" );
1304             secondary_features_map_files = new File[ secondary_features_map_files_strs.length ];
1305             domain_id_to_secondary_features_maps = new Map[ secondary_features_map_files_strs.length ];
1306             int i = 0;
1307             for( final String secondary_features_map_files_str : secondary_features_map_files_strs ) {
1308                 secondary_features_map_files[ i ] = new File( secondary_features_map_files_str );
1309                 final String error = ForesterUtil.isReadableFile( secondary_features_map_files[ i ] );
1310                 if ( !ForesterUtil.isEmpty( error ) ) {
1311                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read secondary features map file: "
1312                             + error );
1313                 }
1314                 try {
1315                     domain_id_to_secondary_features_maps[ i ] = SurfacingUtil
1316                             .createDomainIdToSecondaryFeaturesMap( secondary_features_map_files[ i ] );
1317                 }
1318                 catch ( final IOException e ) {
1319                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "cannot read secondary features map file: "
1320                             + e.getMessage() );
1321                 }
1322                 catch ( final Exception e ) {
1323                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "problem with contents of features map file ["
1324                             + secondary_features_map_files[ i ] + "]: " + e.getMessage() );
1325                 }
1326                 i++;
1327             }
1328         }
1329         if ( out_dir == null ) {
1330             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no output directory indicated (-"
1331                     + surfacing_hmmpfam.OUTPUT_DIR_OPTION + "=<dir>)" );
1332         }
1333         if ( output_file == null ) {
1334             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no name for (main) output file indicated (-"
1335                     + surfacing_hmmpfam.OUTPUT_FILE_OPTION + "=<file>)" );
1336         }
1337         if ( ( domain_id_to_go_ids_map == null ) || domain_id_to_go_ids_map.isEmpty() ) {
1338             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
1339                                      "no (acceptable) Pfam to GO id mapping file provided ('pfam2go file') (-"
1340                                              + surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION + "=<file>)" );
1341         }
1342         if ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) {
1343             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
1344                                      "no (acceptable) go id to term mapping file provided ('GO OBO file') (-"
1345                                              + surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION + "=<file>)" );
1346         }
1347         boolean display_histograms = false;
1348         if ( cla.isOptionSet( surfacing_hmmpfam.DISPLAY_M_HISTOGRAMS_OPTION ) ) {
1349             display_histograms = true;
1350         }
1351         System.out.println( "Output directory            : " + out_dir );
1352         if ( input_file_names_from_file != null ) {
1353             System.out.println( "Input files names from      : " + input_files_file + " ["
1354                     + input_file_names_from_file.length + " input files]" );
1355             html_desc.append( "<tr><td>Input files names from:</td><td>" + input_files_file + " ["
1356                     + input_file_names_from_file.length + " input files]</td></tr>" + nl );
1357         }
1358         if ( positive_filter_file != null ) {
1359             final int filter_size = filter.size();
1360             System.out.println( "Positive protein filter     : " + positive_filter_file + " [" + filter_size
1361                     + " domain ids]" );
1362             html_desc.append( "<tr><td>Positive protein filter:</td><td>" + positive_filter_file + " [" + filter_size
1363                     + " domain ids]</td></tr>" + nl );
1364         }
1365         if ( negative_filter_file != null ) {
1366             final int filter_size = filter.size();
1367             System.out.println( "Negative protein filter     : " + negative_filter_file + " [" + filter_size
1368                     + " domain ids]" );
1369             html_desc.append( "<tr><td>Negative protein filter:</td><td>" + negative_filter_file + " [" + filter_size
1370                     + " domain ids]</td></tr>" + nl );
1371         }
1372         if ( negative_domains_filter_file != null ) {
1373             final int filter_size = filter.size();
1374             System.out.println( "Negative domain filter      : " + negative_domains_filter_file + " [" + filter_size
1375                     + " domain ids]" );
1376             html_desc.append( "<tr><td>Negative domain filter:</td><td>" + negative_domains_filter_file + " ["
1377                     + filter_size + " domain ids]</td></tr>" + nl );
1378         }
1379         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
1380             String plus0 = "";
1381             for( final String s : plus_minus_analysis_high_copy_base_species ) {
1382                 plus0 += "+" + s + " ";
1383             }
1384             String plus1 = "";
1385             for( final String s : plus_minus_analysis_high_copy_target_species ) {
1386                 plus1 += "*" + s + " ";
1387             }
1388             String minus = "";
1389             for( final String s : plus_minus_analysis_high_low_copy_species ) {
1390                 minus += "-" + s + " ";
1391             }
1392             System.out.println( "Plus-minus analysis         : " + plus1 + "&& " + plus0 + "&& " + minus );
1393             html_desc.append( "<tr><td>Plus-minus analysis:</td><td>" + plus1 + "&& " + plus0 + "&& " + minus
1394                     + "</td></tr>" + nl );
1395         }
1396         if ( cutoff_scores_file != null ) {
1397             System.out.println( "Cutoff scores file          : " + cutoff_scores_file );
1398             html_desc.append( "<tr><td>Cutoff scores file:</td><td>" + cutoff_scores_file + "</td></tr>" + nl );
1399         }
1400         if ( e_value_max >= 0.0 ) {
1401             System.out.println( "E-value maximum (inclusive) : " + e_value_max );
1402             html_desc.append( "<tr><td>E-value maximum (inclusive):</td><td>" + e_value_max + "</td></tr>" + nl );
1403         }
1404         System.out.println( "Ignore DUFs                 : " + ignore_dufs );
1405         if ( ignore_virus_like_ids ) {
1406             System.out.println( "Ignore virus like ids       : " + ignore_virus_like_ids );
1407             html_desc.append( "<tr><td>Ignore virus, phage, transposition related ids:</td><td>"
1408                     + ignore_virus_like_ids + "</td></tr>" + nl );
1409         }
1410         html_desc.append( "<tr><td>Ignore DUFs:</td><td>" + ignore_dufs + "</td></tr>" + nl );
1411         if ( max_allowed_overlap != surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1412             System.out.println( "Max allowed domain overlap  : " + max_allowed_overlap );
1413             html_desc.append( "<tr><td>Max allowed domain overlap:</td><td>" + max_allowed_overlap + "</td></tr>" + nl );
1414         }
1415         if ( no_engulfing_overlaps ) {
1416             System.out.println( "Ignore engulfed domains     : " + no_engulfing_overlaps );
1417             html_desc.append( "<tr><td>Ignore (lower confidence) engulfed domains:</td><td>" + no_engulfing_overlaps
1418                     + "</td></tr>" + nl );
1419         }
1420         System.out.println( "Ignore singlet domains      : " + ignore_domains_without_combs_in_all_spec );
1421         html_desc
1422                 .append( "<tr><td>Ignore singlet domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1423                         + ignore_domains_without_combs_in_all_spec + "</td></tr>" + nl );
1424         System.out.println( "Ignore species specific doms: " + ignore_species_specific_domains );
1425         html_desc
1426                 .append( "<tr><td>Ignore species specific domains for domain combination similarity analyses (not for parsimony analyses):</td><td>"
1427                         + ignore_species_specific_domains + "</td></tr>" + nl );
1428         System.out.println( "Ignore combination with self: " + ignore_combination_with_same );
1429         html_desc.append( "<tr><td>Ignore combination with self for domain combination similarity analyses:</td><td>"
1430                 + ignore_combination_with_same + "</td></tr>" + nl );
1431         ;
1432         System.out.println( "Consider directedness       : "
1433                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) );
1434         html_desc.append( "<tr><td>Consider directedness of binary domain combinations:</td><td>"
1435                 + ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) + "</td></tr>" + nl );
1436         if ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) {
1437             System.out.println( "Consider adjacency          : "
1438                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) );
1439             html_desc.append( "<tr><td>Consider djacency of binary domain combinations:</td><td>"
1440                     + ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) + "</td></tr>"
1441                     + nl );
1442         }
1443         System.out.print( "Domain counts sort order    : " );
1444         switch ( dc_sort_order ) {
1445             case ALPHABETICAL_KEY_ID:
1446                 System.out.println( "alphabetical" );
1447                 break;
1448             case KEY_DOMAIN_COUNT:
1449                 System.out.println( "domain count" );
1450                 break;
1451             case KEY_DOMAIN_PROTEINS_COUNT:
1452                 System.out.println( "domain proteins count" );
1453                 break;
1454             case COMBINATIONS_COUNT:
1455                 System.out.println( "domain combinations count" );
1456                 break;
1457             default:
1458                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME, "unknown value for dc sort order" );
1459         }
1460         if ( domain_id_to_go_ids_map != null ) {
1461             System.out.println( "Pfam to GO mappings from    : " + pfam_to_go_file + " [" + domain_id_to_go_ids_count
1462                     + " mappings]" );
1463             html_desc.append( "<tr><td>Pfam to GO mappings from:</td><td>" + pfam_to_go_file + " ["
1464                     + domain_id_to_go_ids_count + " mappings]" + "</td></tr>" + nl );
1465         }
1466         if ( go_terms != null ) {
1467             System.out.println( "GO terms from               : " + go_obo_file + " [" + go_terms.size() + " terms]" );
1468             html_desc.append( "<tr><td>GO terms from:</td><td>" + go_obo_file + " [" + go_terms.size() + " terms]"
1469                     + "</td></tr>" + nl );
1470         }
1471         if ( go_namespace_limit != null ) {
1472             System.out.println( "Limit GO terms to           : " + go_namespace_limit.toString() );
1473             html_desc.append( "<tr><td>Limit GO terms to</td><td>" + go_namespace_limit + "</td></tr>" + nl );
1474         }
1475         if ( perform_pwc ) {
1476             System.out.println( "Suffix for PWC files        : " + automated_pairwise_comparison_suffix );
1477             html_desc.append( "<tr><td>Suffix for PWC files</td><td>" + automated_pairwise_comparison_suffix
1478                     + "</td></tr>" + nl );
1479         }
1480         if ( out_dir != null ) {
1481             System.out.println( "Output directory            : " + out_dir );
1482         }
1483         if ( query_domain_ids != null ) {
1484             System.out.println( "Query domains (ordered)     : " + query_domain_ids );
1485             html_desc.append( "<tr><td></td><td>" + query_domain_ids + "</td></tr>" + nl );
1486         }
1487         System.out.println( "Write similarities to       : " + output_file );
1488         System.out.print( "  Scoring method            : " );
1489         html_desc.append( "<tr><td>Scoring method:</td><td>" );
1490         switch ( scoring ) {
1491             case COMBINATIONS:
1492                 System.out.println( "domain combinations based" );
1493                 html_desc.append( "domain combinations based" + "</td></tr>" + nl );
1494                 break;
1495             case DOMAINS:
1496                 System.out.println( "domain counts based" );
1497                 html_desc.append( "domain counts based" + "</td></tr>" + nl );
1498                 break;
1499             case PROTEINS:
1500                 System.out.println( "domain proteins counts based" );
1501                 html_desc.append( "domain proteins counts based" + "</td></tr>" + nl );
1502                 break;
1503             default:
1504                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME, "unknown value for sorting for scoring" );
1505         }
1506         System.out.print( "  Sort by                   : " );
1507         html_desc.append( "<tr><td>Sort by:</td><td>" );
1508         switch ( domain_similarity_sort_field ) {
1509             case MIN:
1510                 System.out.print( "score minimum" );
1511                 html_desc.append( "score minimum" );
1512                 break;
1513             case MAX:
1514                 System.out.print( "score maximum" );
1515                 html_desc.append( "score maximum" );
1516                 break;
1517             case MEAN:
1518                 System.out.print( "score mean" );
1519                 html_desc.append( "score mean" );
1520                 break;
1521             case SD:
1522                 System.out.print( "score standard deviation" );
1523                 html_desc.append( "score standard deviation" );
1524                 break;
1525             case SPECIES_COUNT:
1526                 System.out.print( "species number" );
1527                 html_desc.append( "species number" );
1528                 break;
1529             case DOMAIN_ID:
1530                 System.out.print( "alphabetical domain identifier" );
1531                 html_desc.append( "alphabetical domain identifier" );
1532                 break;
1533             case MAX_DIFFERENCE:
1534                 System.out.print( "(maximal) difference" );
1535                 html_desc.append( "(maximal) difference" );
1536                 break;
1537             case ABS_MAX_COUNTS_DIFFERENCE:
1538                 System.out.print( "absolute (maximal) counts difference" );
1539                 html_desc.append( "absolute (maximal) counts difference" );
1540                 break;
1541             case MAX_COUNTS_DIFFERENCE:
1542                 System.out.print( "(maximal) counts difference" );
1543                 html_desc.append( "(maximal) counts  difference" );
1544                 break;
1545             default:
1546                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME,
1547                                                    "unknown value for sorting for similarities" );
1548         }
1549         if ( sort_by_species_count_first ) {
1550             System.out.println( " (sort by species count first)" );
1551             html_desc.append( " (sort by species count first)" );
1552         }
1553         else {
1554             System.out.println();
1555         }
1556         html_desc.append( "</td></tr>" + nl );
1557         System.out.print( "  Detailedness              : " );
1558         switch ( detailedness ) {
1559             case BASIC:
1560                 System.out.println( "basic" );
1561                 break;
1562             case LIST_COMBINING_DOMAIN_FOR_EACH_SPECIES:
1563                 System.out.println( "list combining domains for each species" );
1564                 break;
1565             case PUNCTILIOUS:
1566                 System.out.println( "punctilious" );
1567                 break;
1568             default:
1569                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME,
1570                                                    "unknown value for sorting for detailedness" );
1571         }
1572         System.out.print( "  Print option              : " );
1573         switch ( domain_similarity_print_option ) {
1574             case HTML:
1575                 System.out.println( "HTML" );
1576                 break;
1577             case SIMPLE_TAB_DELIMITED:
1578                 System.out.println( "simple tab delimited" );
1579                 break;
1580             default:
1581                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME, "unknown value for print option" );
1582         }
1583         System.out.print( "  Species matrix            : " + species_matrix );
1584         System.out.println();
1585         if ( perform_pwc ) {
1586             System.out.println( "Pairwise comparisons: " );
1587             html_desc.append( "<tr><td>Pairwise comparisons:</td><td></td></tr>" );
1588             System.out.print( "  Sort by                   : " );
1589             html_desc.append( "<tr><td>Sort by:</td><td>" );
1590             switch ( domain_similarity_sort_field_for_automated_pwc ) {
1591                 case MEAN:
1592                     System.out.print( "score mean" );
1593                     html_desc.append( "score mean" );
1594                     break;
1595                 case DOMAIN_ID:
1596                     System.out.print( "alphabetical domain identifier" );
1597                     html_desc.append( "alphabetical domain identifier" );
1598                     break;
1599                 case MAX_DIFFERENCE:
1600                     System.out.print( "difference" );
1601                     html_desc.append( "difference" );
1602                     break;
1603                 case ABS_MAX_COUNTS_DIFFERENCE:
1604                     System.out.print( "absolute counts difference" );
1605                     html_desc.append( "absolute counts difference" );
1606                     break;
1607                 case MAX_COUNTS_DIFFERENCE:
1608                     System.out.print( "counts difference" );
1609                     html_desc.append( "counts difference" );
1610                     break;
1611                 default:
1612                     ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME,
1613                                                        "unknown value for sorting for similarities" );
1614             }
1615             System.out.println();
1616             html_desc.append( "</td></tr>" + nl );
1617             if ( jacknifed_distances ) {
1618                 html_desc.append( "<tr><td>Jacknife:</td><td>" + jacknife_resamplings + " resamplings</td></tr>" + nl );
1619                 html_desc.append( "<tr><td>Jacknife ratio:</td><td>" + ForesterUtil.round( jacknife_ratio, 2 )
1620                         + "</td></tr>" + nl );
1621                 html_desc.append( "<tr><td>Jacknife random number seed:</td><td>" + random_seed + "</td></tr>" + nl );
1622                 System.out.println( "  Jacknife                  : " + jacknife_resamplings + " resamplings" );
1623                 System.out.println( "    Ratio                   : " + ForesterUtil.round( jacknife_ratio, 2 ) );
1624                 System.out.println( "    Random number seed      : " + random_seed );
1625             }
1626             //                if ( infer_species_trees ) {
1627             //                    html_desc.append( "<tr><td>Infer species trees:</td><td>true</td></tr>" + nl );
1628             //                    System.out.println( "  Infer species trees       : true" );
1629             //                }
1630             if ( ( intrees != null ) && ( intrees.length > 0 ) ) {
1631                 for( final File intree_file : intree_files ) {
1632                     html_desc.append( "<tr><td>Intree for gain/loss parsimony analysis:</td><td>" + intree_file
1633                             + "</td></tr>" + nl );
1634                     System.out.println( "  Intree for gain/loss pars.: " + intree_file );
1635                 }
1636             }
1637             if ( radomize_fitch_parsimony ) {
1638                 html_desc.append( "<tr><td>    Random number seed for Fitch parsimony analysis:</td><td>"
1639                         + random_number_seed_for_fitch_parsimony + "</td></tr>" + nl );
1640                 System.out.println( "    Random number seed      : " + random_number_seed_for_fitch_parsimony );
1641             }
1642             if ( ( domain_id_to_secondary_features_maps != null ) && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
1643                 for( int i = 0; i < secondary_features_map_files.length; i++ ) {
1644                     html_desc.append( "<tr><td>Secondary features map file:</td><td>"
1645                             + secondary_features_map_files[ i ] + "</td></tr>" + nl );
1646                     System.out.println( "Secondary features map file : " + secondary_features_map_files[ i ]
1647                             + " [mappings for " + domain_id_to_secondary_features_maps[ i ].size() + " domain ids]" );
1648                     if ( VERBOSE ) {
1649                         System.out.println();
1650                         System.out.println( "Domain ids to secondary features map:" );
1651                         for( final DomainId domain_id : domain_id_to_secondary_features_maps[ i ].keySet() ) {
1652                             System.out.print( domain_id.getId() );
1653                             System.out.print( " => " );
1654                             for( final String sec : domain_id_to_secondary_features_maps[ i ].get( domain_id ) ) {
1655                                 System.out.print( sec );
1656                                 System.out.print( " " );
1657                             }
1658                             System.out.println();
1659                         }
1660                     }
1661                 }
1662             }
1663         } // if ( perform_pwc ) {
1664         System.out.println();
1665         html_desc.append( "<tr><td>Command line:</td><td>" + cla.getCommandLineArgsAsString() + "</td></tr>" + nl );
1666         System.out.println( "Command line                : " + cla.getCommandLineArgsAsString() );
1667         BufferedWriter[] query_domains_writer_ary = null;
1668         List<DomainId>[] query_domain_ids_array = null;
1669         if ( query_domain_ids != null ) {
1670             final String[] query_domain_ids_str_array = query_domain_ids.split( "#" );
1671             query_domain_ids_array = new ArrayList[ query_domain_ids_str_array.length ];
1672             query_domains_writer_ary = new BufferedWriter[ query_domain_ids_str_array.length ];
1673             for( int i = 0; i < query_domain_ids_str_array.length; i++ ) {
1674                 String query_domain_ids_str = query_domain_ids_str_array[ i ];
1675                 final String[] query_domain_ids_str_ary = query_domain_ids_str.split( "~" );
1676                 final List<DomainId> query = new ArrayList<DomainId>();
1677                 for( final String element : query_domain_ids_str_ary ) {
1678                     query.add( new DomainId( element ) );
1679                 }
1680                 query_domain_ids_array[ i ] = query;
1681                 query_domain_ids_str = query_domain_ids_str.replace( '~', '_' );
1682                 String protein_names_writer_str = query_domain_ids_str + surfacing_hmmpfam.SEQ_EXTRACT_SUFFIX;
1683                 if ( out_dir != null ) {
1684                     protein_names_writer_str = out_dir + ForesterUtil.FILE_SEPARATOR + protein_names_writer_str;
1685                 }
1686                 try {
1687                     query_domains_writer_ary[ i ] = new BufferedWriter( new FileWriter( protein_names_writer_str ) );
1688                 }
1689                 catch ( final IOException e ) {
1690                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "Could not open [" + protein_names_writer_str
1691                             + "]: " + e.getLocalizedMessage() );
1692                 }
1693             }
1694         }
1695         SortedMap<Species, List<Protein>> protein_lists_per_species = null; //This will only be created if neede.
1696         boolean need_protein_lists_per_species = false;
1697         if ( ( plus_minus_analysis_high_copy_base_species.size() > 0 ) || output_protein_lists_for_all_domains ) {
1698             need_protein_lists_per_species = true;
1699         }
1700         if ( need_protein_lists_per_species ) {
1701             protein_lists_per_species = new TreeMap<Species, List<Protein>>();
1702         }
1703         final List<GenomeWideCombinableDomains> gwcd_list = new ArrayList<GenomeWideCombinableDomains>( number_of_genomes );
1704         final SortedSet<DomainId> all_domains_encountered = new TreeSet<DomainId>();
1705         final SortedSet<BinaryDomainCombination> all_bin_domain_combinations_encountered = new TreeSet<BinaryDomainCombination>();
1706         List<BinaryDomainCombination> all_bin_domain_combinations_gained_fitch = null;
1707         List<BinaryDomainCombination> all_bin_domain_combinations_lost_fitch = null;
1708         if ( ( intrees != null ) && ( intrees.length == 1 ) ) {
1709             all_bin_domain_combinations_gained_fitch = new ArrayList<BinaryDomainCombination>();
1710             all_bin_domain_combinations_lost_fitch = new ArrayList<BinaryDomainCombination>();
1711         }
1712         final DomainLengthsTable domain_lengths_table = new DomainLengthsTable();
1713         final File per_genome_domain_promiscuity_statistics_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR
1714                 + output_file + D_PROMISCUITY_FILE_SUFFIX );
1715         BufferedWriter per_genome_domain_promiscuity_statistics_writer = null;
1716         try {
1717             per_genome_domain_promiscuity_statistics_writer = new BufferedWriter( new FileWriter( per_genome_domain_promiscuity_statistics_file ) );
1718             per_genome_domain_promiscuity_statistics_writer.write( "Species:\t" );
1719             per_genome_domain_promiscuity_statistics_writer.write( "Mean:\t" );
1720             per_genome_domain_promiscuity_statistics_writer.write( "SD:\t" );
1721             per_genome_domain_promiscuity_statistics_writer.write( "Median:\t" );
1722             per_genome_domain_promiscuity_statistics_writer.write( "Min:\t" );
1723             per_genome_domain_promiscuity_statistics_writer.write( "Max:\t" );
1724             per_genome_domain_promiscuity_statistics_writer.write( "N:\t" );
1725             per_genome_domain_promiscuity_statistics_writer.write( "Max Promiscuous Domains:"
1726                     + ForesterUtil.LINE_SEPARATOR );
1727         }
1728         catch ( final IOException e2 ) {
1729             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e2.getMessage() );
1730         }
1731         for( int i = 0; i < number_of_genomes; ++i ) {
1732             System.out.println();
1733             System.out.println( ( i + 1 ) + "/" + number_of_genomes );
1734             System.out.println( "Processing                                     : " + input_file_properties[ i ][ 0 ] );
1735             HmmPfamOutputParser parser = null;
1736             if ( ( positive_filter_file != null ) || ( negative_filter_file != null )
1737                     || ( negative_domains_filter_file != null ) ) {
1738                 HmmPfamOutputParser.FilterType filter_type = HmmPfamOutputParser.FilterType.NONE;
1739                 if ( positive_filter_file != null ) {
1740                     filter_type = HmmPfamOutputParser.FilterType.POSITIVE_PROTEIN;
1741                 }
1742                 else if ( negative_filter_file != null ) {
1743                     filter_type = HmmPfamOutputParser.FilterType.NEGATIVE_PROTEIN;
1744                 }
1745                 else if ( negative_domains_filter_file != null ) {
1746                     filter_type = HmmPfamOutputParser.FilterType.NEGATIVE_DOMAIN;
1747                 }
1748                 parser = new HmmPfamOutputParser( new File( input_file_properties[ i ][ 0 ] ),
1749                                                   input_file_properties[ i ][ 1 ],
1750                                                   input_file_properties[ i ][ 2 ],
1751                                                   filter,
1752                                                   filter_type );
1753             }
1754             else {
1755                 parser = new HmmPfamOutputParser( new File( input_file_properties[ i ][ 0 ] ),
1756                                                   input_file_properties[ i ][ 1 ],
1757                                                   input_file_properties[ i ][ 2 ] );
1758             }
1759             if ( e_value_max >= 0.0 ) {
1760                 parser.setEValueMaximum( e_value_max );
1761             }
1762             parser.setIgnoreDufs( ignore_dufs );
1763             parser.setIgnoreVirusLikeIds( ignore_virus_like_ids );
1764             parser.setIgnoreEngulfedDomains( no_engulfing_overlaps );
1765             if ( max_allowed_overlap != surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_DEFAULT ) {
1766                 parser.setMaxAllowedOverlap( max_allowed_overlap );
1767             }
1768             parser.setReturnType( HmmPfamOutputParser.ReturnType.UNORDERED_PROTEIN_DOMAIN_COLLECTION_PER_PROTEIN );
1769             if ( individual_domain_score_cutoffs != null ) {
1770                 parser.setIndividualDomainScoreCutoffs( individual_domain_score_cutoffs );
1771             }
1772             parser.setAllowNonUniqueQuery( ALLOW_NON_UNIQUE_QUERY_IN_HMMPFAM_OUTPUT_DEFAULT );
1773             parser.setVerbose( VERBOSE_DEFAULT );
1774             List<Protein> protein_list = null;
1775             try {
1776                 protein_list = parser.parse();
1777             }
1778             catch ( final IOException e ) {
1779                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage() );
1780             }
1781             catch ( final Exception e ) {
1782                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage(), e );
1783             }
1784             if ( VERBOSE ) {
1785                 System.out.println( "Domains ignored due to negative domain filter: " );
1786                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToNegativeDomainFilterCountsMap() );
1787                 System.out.println( "Domains ignored due to virus like id: " );
1788                 ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToVirusLikeIdCountsMap() );
1789             }
1790             System.out.println( "Number of proteins encountered                 : " + parser.getProteinsEncountered() );
1791             System.out.println( "Number of proteins stored                      : " + protein_list.size() );
1792             System.out.println( "Domains encountered                            : " + parser.getDomainsEncountered() );
1793             System.out.println( "Domains stored                                 : " + parser.getDomainsStored() );
1794             System.out.println( "Distinct domains stored                        : "
1795                     + parser.getDomainsStoredSet().size() );
1796             System.out.println( "Domains ignored due to individual score cutoffs: "
1797                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff() );
1798             System.out.println( "Domains ignored due to E-value                 : "
1799                     + parser.getDomainsIgnoredDueToEval() );
1800             System.out.println( "Domains ignored due to DUF designation         : "
1801                     + parser.getDomainsIgnoredDueToDuf() );
1802             if ( ignore_virus_like_ids ) {
1803                 System.out.println( "Domains ignored due virus like ids             : "
1804                         + parser.getDomainsIgnoredDueToVirusLikeIds() );
1805             }
1806             System.out.println( "Domains ignored due negative domain filter     : "
1807                     + parser.getDomainsIgnoredDueToNegativeDomainFilter() );
1808             System.out.println( "Domains ignored due to overlap                 : "
1809                     + parser.getDomainsIgnoredDueToOverlap() );
1810             if ( negative_filter_file != null ) {
1811                 System.out.println( "Proteins ignored due to negative filter        : "
1812                         + parser.getProteinsIgnoredDueToFilter() );
1813             }
1814             if ( positive_filter_file != null ) {
1815                 System.out.println( "Proteins ignored due to positive filter        : "
1816                         + parser.getProteinsIgnoredDueToFilter() );
1817             }
1818             System.out.println( "Time for processing                            : " + parser.getTime() + "ms" );
1819             html_desc.append( "<tr><td>" + input_file_properties[ i ][ 0 ] + " [species: "
1820                     + input_file_properties[ i ][ 1 ] + "]" + ":</td><td>domains analyzed: "
1821                     + parser.getDomainsStored() + "; domains ignored: [ind score cutoffs: "
1822                     + parser.getDomainsIgnoredDueToIndividualScoreCutoff() + "] [E-value cutoff: "
1823                     + parser.getDomainsIgnoredDueToEval() + "] [DUF: " + parser.getDomainsIgnoredDueToDuf()
1824                     + "] [virus like ids: " + parser.getDomainsIgnoredDueToVirusLikeIds()
1825                     + "] [negative domain filter: " + parser.getDomainsIgnoredDueToNegativeDomainFilter()
1826                     + "] [overlap: " + parser.getDomainsIgnoredDueToOverlap() + "]" );
1827             if ( negative_filter_file != null ) {
1828                 html_desc.append( "; proteins ignored due to negative filter: "
1829                         + parser.getProteinsIgnoredDueToFilter() );
1830             }
1831             if ( positive_filter_file != null ) {
1832                 html_desc.append( "; proteins ignored due to positive filter: "
1833                         + parser.getProteinsIgnoredDueToFilter() );
1834             }
1835             html_desc.append( "</td></tr>" + nl );
1836             // domain_partner_counts_array[ i ] =
1837             // Methods.getDomainPartnerCounts( protein_domain_collections_array[
1838             // i ],
1839             // false, input_file_properties[ i ][ 1 ] );
1840             gwcd_list.add( BasicGenomeWideCombinableDomains
1841                     .createInstance( protein_list,
1842                                      ignore_combination_with_same,
1843                                      new BasicSpecies( input_file_properties[ i ][ 1 ] ),
1844                                      domain_id_to_go_ids_map,
1845                                      dc_type ) );
1846             domain_lengths_table.addLengths( protein_list );
1847             if ( gwcd_list.get( i ).getSize() > 0 ) {
1848                 SurfacingUtil.writeDomainCombinationsCountsFile( input_file_properties,
1849                                                                  out_dir,
1850                                                                  per_genome_domain_promiscuity_statistics_writer,
1851                                                                  gwcd_list.get( i ),
1852                                                                  i,
1853                                                                  dc_sort_order );
1854                 if ( output_binary_domain_combinationsfor_graph_analysis ) {
1855                     SurfacingUtil.writeBinaryDomainCombinationsFileForGraphAnalysis( input_file_properties,
1856                                                                                      out_dir,
1857                                                                                      gwcd_list.get( i ),
1858                                                                                      i,
1859                                                                                      dc_sort_order );
1860                 }
1861                 SurfacingUtil.addAllDomainIdsToSet( gwcd_list.get( i ), all_domains_encountered );
1862                 SurfacingUtil.addAllBinaryDomainCombinationToSet( gwcd_list.get( i ),
1863                                                                   all_bin_domain_combinations_encountered );
1864             }
1865             if ( query_domains_writer_ary != null ) {
1866                 for( int j = 0; j < query_domain_ids_array.length; j++ ) {
1867                     try {
1868                         SurfacingUtil.extractProteinNames( protein_list,
1869                                                            query_domain_ids_array[ j ],
1870                                                            query_domains_writer_ary[ j ],
1871                                                            "\t",
1872                                                            surfacing.LIMIT_SPEC_FOR_PROT_EX );
1873                         query_domains_writer_ary[ j ].flush();
1874                     }
1875                     catch ( final IOException e ) {
1876                         e.printStackTrace();
1877                     }
1878                 }
1879             }
1880             if ( need_protein_lists_per_species ) {
1881                 protein_lists_per_species.put( new BasicSpecies( input_file_properties[ i ][ 1 ] ), protein_list );
1882             }
1883             System.gc();
1884         } // for( int i = 0; i < number_of_hmmpfam_files_to_analyze; ++i ) {
1885         try {
1886             per_genome_domain_promiscuity_statistics_writer.flush();
1887             per_genome_domain_promiscuity_statistics_writer.close();
1888         }
1889         catch ( final IOException e2 ) {
1890             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e2.toString() );
1891         }
1892         ForesterUtil.programMessage( PRG_NAME, "Wrote domain promiscuities to: "
1893                 + per_genome_domain_promiscuity_statistics_file );
1894         if ( query_domains_writer_ary != null ) {
1895             for( int j = 0; j < query_domain_ids_array.length; j++ ) {
1896                 try {
1897                     query_domains_writer_ary[ j ].close();
1898                 }
1899                 catch ( final IOException e ) {
1900                     ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.toString() );
1901                 }
1902             }
1903         }
1904         if ( PERFORM_DOMAIN_LENGTH_ANALYSIS ) {
1905             try {
1906                 SurfacingUtil.executeDomainLengthAnalysis( input_file_properties,
1907                                                            number_of_genomes,
1908                                                            domain_lengths_table,
1909                                                            domain_lengths_analysis_outfile );
1910             }
1911             catch ( final IOException e1 ) {
1912                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e1.toString() );
1913             }
1914             System.out.println();
1915             ForesterUtil.programMessage( PRG_NAME, "Wrote domain length data to: " + domain_lengths_analysis_outfile );
1916             System.out.println();
1917         }
1918         final long analysis_start_time = new Date().getTime();
1919         PairwiseDomainSimilarityCalculator pw_calc = null;
1920         // double[] values_for_all_scores_histogram = null;
1921         final DomainSimilarityCalculator calc = new BasicDomainSimilarityCalculator( domain_similarity_sort_field,
1922                                                                                      sort_by_species_count_first,
1923                                                                                      number_of_genomes == 2 );
1924         switch ( scoring ) {
1925             case COMBINATIONS:
1926                 pw_calc = new CombinationsBasedPairwiseDomainSimilarityCalculator();
1927                 break;
1928             case DOMAINS:
1929                 pw_calc = new DomainCountsBasedPairwiseSimilarityCalculator();
1930                 break;
1931             case PROTEINS:
1932                 pw_calc = new ProteinCountsBasedPairwiseDomainSimilarityCalculator();
1933                 break;
1934             default:
1935                 ForesterUtil.unexpectedFatalError( surfacing_hmmpfam.PRG_NAME, "unknown value for sorting for scoring" );
1936         }
1937         DomainSimilarityCalculator.GoAnnotationOutput go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.NONE;
1938         if ( domain_id_to_go_ids_map != null ) {
1939             go_annotation_output = DomainSimilarityCalculator.GoAnnotationOutput.ALL;
1940         }
1941         final SortedSet<DomainSimilarity> similarities = calc
1942                 .calculateSimilarities( pw_calc,
1943                                         gwcd_list,
1944                                         ignore_domains_without_combs_in_all_spec,
1945                                         ignore_species_specific_domains );
1946         SurfacingUtil.decoratePrintableDomainSimilarities( similarities,
1947                                                            detailedness,
1948                                                            go_annotation_output,
1949                                                            go_id_to_term_map,
1950                                                            go_namespace_limit );
1951         DescriptiveStatistics pw_stats = null;
1952         try {
1953             String my_outfile = output_file.toString();
1954             if ( !my_outfile.endsWith( ".html" ) ) {
1955                 my_outfile += ".html";
1956             }
1957             final Writer writer = new BufferedWriter( new FileWriter( out_dir == null ? my_outfile : out_dir
1958                     + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
1959             List<Species> species_order = null;
1960             if ( species_matrix ) {
1961                 species_order = new ArrayList<Species>();
1962                 for( int i = 0; i < number_of_genomes; i++ ) {
1963                     species_order.add( new BasicSpecies( input_file_properties[ i ][ 1 ] ) );
1964                 }
1965             }
1966             html_desc.append( "<tr><td>Sum of all distinct binary combinations:</td><td>"
1967                     + all_bin_domain_combinations_encountered.size() + "</td></tr>" + nl );
1968             html_desc.append( "<tr><td>Sum of all distinct domains:</td><td>" + all_domains_encountered.size()
1969                     + "</td></tr>" + nl );
1970             html_desc.append( "<tr><td>Analysis date/time:</td><td>"
1971                     + new java.text.SimpleDateFormat( "yyyy.MM.dd HH:mm:ss" ).format( new java.util.Date() )
1972                     + "</td></tr>" + nl );
1973             html_desc.append( "</table>" + nl );
1974             pw_stats = SurfacingUtil
1975                     .writeDomainSimilaritiesToFile( html_desc,
1976                                                     new StringBuilder( number_of_genomes + " genomes" ),
1977                                                     writer,
1978                                                     null,
1979                                                     similarities,
1980                                                     number_of_genomes == 2,
1981                                                     species_order,
1982                                                     domain_similarity_print_option,
1983                                                     domain_similarity_sort_field,
1984                                                     scoring,
1985                                                     true );
1986             ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME,
1987                                          "Wrote main output (includes domain similarities) to: \""
1988                                                  + ( out_dir == null ? my_outfile : out_dir
1989                                                          + ForesterUtil.FILE_SEPARATOR + my_outfile ) + "\"" );
1990         }
1991         catch ( final IOException e ) {
1992             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "Failed to write similarites to: \"" + output_file
1993                     + "\" [" + e.getMessage() + "]" );
1994         }
1995         System.out.println();
1996         // values_for_all_scores_histogram = pw_stats.getDataAsDoubleArray();
1997         final Species[] species = new Species[ number_of_genomes ];
1998         for( int i = 0; i < number_of_genomes; ++i ) {
1999             species[ i ] = new BasicSpecies( input_file_properties[ i ][ 1 ] );
2000         }
2001         List<Phylogeny> inferred_trees = null;
2002         if ( ( number_of_genomes > 2 ) && perform_pwc ) {
2003             final PairwiseGenomeComparator pwgc = new PairwiseGenomeComparator();
2004             pwgc.performPairwiseComparisons( html_desc,
2005                                              sort_by_species_count_first,
2006                                              detailedness,
2007                                              ignore_domains_without_combs_in_all_spec,
2008                                              ignore_species_specific_domains,
2009                                              domain_similarity_sort_field_for_automated_pwc,
2010                                              domain_similarity_print_option,
2011                                              scoring,
2012                                              domain_id_to_go_ids_map,
2013                                              go_id_to_term_map,
2014                                              go_namespace_limit,
2015                                              species,
2016                                              number_of_genomes,
2017                                              gwcd_list,
2018                                              pw_calc,
2019                                              automated_pairwise_comparison_suffix,
2020                                              true,
2021                                              surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_PREFIX,
2022                                              surfacing_hmmpfam.PRG_NAME,
2023                                              out_dir,
2024                                              write_pwc_files );
2025             String matrix_output_file = new String( output_file.toString() );
2026             if ( matrix_output_file.indexOf( '.' ) > 1 ) {
2027                 matrix_output_file = matrix_output_file.substring( 0, matrix_output_file.indexOf( '.' ) );
2028             }
2029             if ( out_dir != null ) {
2030                 matrix_output_file = out_dir + ForesterUtil.FILE_SEPARATOR + matrix_output_file;
2031                 output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
2032             }
2033             SurfacingUtil
2034                     .writeMatrixToFile( new File( matrix_output_file
2035                                                 + surfacing_hmmpfam.MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ),
2036                                         pwgc.getDomainDistanceScoresMeans() );
2037             SurfacingUtil
2038                     .writeMatrixToFile( new File( matrix_output_file
2039                                                 + surfacing_hmmpfam.MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ),
2040                                         pwgc.getSharedBinaryCombinationsBasedDistances() );
2041             SurfacingUtil
2042                     .writeMatrixToFile( new File( matrix_output_file
2043                                                 + surfacing_hmmpfam.MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ),
2044                                         pwgc.getSharedDomainsBasedDistances() );
2045             final Phylogeny nj_gd = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2046                     + surfacing_hmmpfam.NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
2047                     .getDomainDistanceScoresMeans().get( 0 ) );
2048             final Phylogeny nj_bc = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2049                     + surfacing_hmmpfam.NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
2050                     .getSharedBinaryCombinationsBasedDistances().get( 0 ) );
2051             final Phylogeny nj_d = SurfacingUtil.createNjTreeBasedOnMatrixToFile( new File( matrix_output_file
2052                     + surfacing_hmmpfam.NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ), pwgc
2053                     .getSharedDomainsBasedDistances().get( 0 ) );
2054             inferred_trees = new ArrayList<Phylogeny>();
2055             inferred_trees.add( nj_gd );
2056             inferred_trees.add( nj_bc );
2057             inferred_trees.add( nj_d );
2058             // final List<HistogramData> histogram_datas = pwgc.getHistogramDatas();
2059             //            if ( infer_species_trees ) {
2060             //                inferred_trees = new ArrayList<Phylogeny>();
2061             //                final List<Phylogeny> inferred_trees_bc =  inferSpeciesTrees( new File( output_file + INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc
2062             //                        .getSharedBinaryCombinationsBasedDistances() );
2063             //                final List<Phylogeny> inferred_trees_d =  inferSpeciesTrees( new File( output_file + INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc
2064             //                        .getSharedDomainsBasedDistances() );
2065             //                inferred_trees.addAll( inferred_trees_bc );
2066             //                inferred_trees.addAll( inferred_trees_d );
2067             //            }
2068             if ( jacknifed_distances ) {
2069                 pwgc.performPairwiseComparisonsJacknifed( species,
2070                                                           number_of_genomes,
2071                                                           gwcd_list,
2072                                                           true,
2073                                                           jacknife_resamplings,
2074                                                           jacknife_ratio,
2075                                                           random_seed );
2076                 SurfacingUtil
2077                         .writeMatrixToFile( new File( matrix_output_file
2078                                                     + "_"
2079                                                     + ForesterUtil.round( jacknife_ratio, 2 )
2080                                                     + "_"
2081                                                     + jacknife_resamplings
2082                                                     + surfacing_hmmpfam.MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX ),
2083                                             pwgc.getSharedBinaryCombinationsBasedDistances() );
2084                 SurfacingUtil
2085                         .writeMatrixToFile( new File( matrix_output_file
2086                                                     + "_"
2087                                                     + ForesterUtil.round( jacknife_ratio, 2 )
2088                                                     + "_"
2089                                                     + jacknife_resamplings
2090                                                     + surfacing_hmmpfam.MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX ),
2091                                             pwgc.getSharedDomainsBasedDistances() );
2092                 //                if ( infer_species_trees ) {
2093                 //                    inferSpeciesTrees( new File( output_file + "_" + jacknife_resamplings
2094                 //                            + INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc
2095                 //                            .getSharedBinaryCombinationsBasedDistances() );
2096                 //                    inferSpeciesTrees( new File( output_file + "_" + jacknife_resamplings
2097                 //                            + INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX ), pwgc.getSharedDomainsBasedDistances() );
2098                 //                }
2099             }
2100             if ( display_histograms ) {
2101                 //                final List<HistogramData> histogram_datas_all = new ArrayList<HistogramData>();
2102                 //                histogram_datas_all.add( new HistogramData( "all",
2103                 //                                                            values_for_all_scores_histogram,
2104                 //                                                            null,
2105                 //                                                            20 ) );
2106                 //                final HistogramsFrame hf_all = new HistogramsFrame( histogram_datas_all );
2107                 //                final HistogramsFrame hf = new HistogramsFrame( histogram_datas );
2108                 //                hf_all.setVisible( true );
2109                 //                hf.setVisible( true );
2110             }
2111         } // if ( ( output_file != null ) && ( number_of_genomes > 2 ) && !isEmpty( automated_pairwise_comparison_suffix ) )
2112         if ( ( out_dir != null ) && ( !perform_pwc ) ) {
2113             output_file = new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file );
2114         }
2115         //    writePresentToNexus( output_file, positive_filter_file, filter, gwcd_list );
2116         if ( ( ( intrees != null ) && ( intrees.length > 0 ) ) && ( number_of_genomes > 2 ) ) {
2117             final StringBuilder parameters_sb = createParametersAsString( ignore_dufs,
2118                                                                           e_value_max,
2119                                                                           max_allowed_overlap,
2120                                                                           no_engulfing_overlaps,
2121                                                                           cutoff_scores_file,
2122                                                                           dc_type );
2123             String s = "_";
2124             if ( radomize_fitch_parsimony ) {
2125                 s += random_number_seed_for_fitch_parsimony + "_";
2126             }
2127             int i = 0;
2128             for( final Phylogeny intree : intrees ) {
2129                 final String outfile_name = ForesterUtil.removeSuffix( output_file.toString() ) + s
2130                         + ForesterUtil.removeSuffix( intree_files[ i ].toString() );
2131                 final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator.createInstance( intree,
2132                                                                                                              gwcd_list );
2133                 SurfacingUtil.executeParsimonyAnalysis( random_number_seed_for_fitch_parsimony,
2134                                                         radomize_fitch_parsimony,
2135                                                         outfile_name,
2136                                                         domain_parsimony,
2137                                                         intree,
2138                                                         domain_id_to_go_ids_map,
2139                                                         go_id_to_term_map,
2140                                                         go_namespace_limit,
2141                                                         parameters_sb.toString(),
2142                                                         domain_id_to_secondary_features_maps,
2143                                                         positive_filter_file == null ? null : filter,
2144                                                         output_binary_domain_combinationsfor_graph_analysis,
2145                                                         all_bin_domain_combinations_gained_fitch,
2146                                                         all_bin_domain_combinations_lost_fitch,
2147                                                         dc_type );
2148                 // Listing of all domain combinations gained is only done if only one input tree is used.
2149                 if ( ( domain_id_to_secondary_features_maps != null )
2150                         && ( domain_id_to_secondary_features_maps.length > 0 ) ) {
2151                     int j = 0;
2152                     for( final Map<DomainId, Set<String>> domain_id_to_secondary_features_map : domain_id_to_secondary_features_maps ) {
2153                         final Map<Species, MappingResults> mapping_results_map = new TreeMap<Species, MappingResults>();
2154                         final DomainParsimonyCalculator secondary_features_parsimony = DomainParsimonyCalculator
2155                                 .createInstance( intree, gwcd_list, domain_id_to_secondary_features_map );
2156                         SurfacingUtil
2157                                 .executeParsimonyAnalysisForSecondaryFeatures( outfile_name
2158                                                                                        + "_"
2159                                                                                        + secondary_features_map_files[ j++ ],
2160                                                                                secondary_features_parsimony,
2161                                                                                intree,
2162                                                                                parameters_sb.toString(),
2163                                                                                mapping_results_map );
2164                         if ( i == 0 ) {
2165                             System.out.println();
2166                             System.out.println( "Mapping to secondary features:" );
2167                             for( final Species spec : mapping_results_map.keySet() ) {
2168                                 final MappingResults mapping_results = mapping_results_map.get( spec );
2169                                 final int total_domains = mapping_results.getSumOfFailures()
2170                                         + mapping_results.getSumOfSuccesses();
2171                                 System.out.print( spec + ":" );
2172                                 System.out.print( " mapped domains = " + mapping_results.getSumOfSuccesses() );
2173                                 System.out.print( ", not mapped domains = " + mapping_results.getSumOfFailures() );
2174                                 if ( total_domains > 0 ) {
2175                                     System.out.println( ", mapped ratio = "
2176                                             + ( 100 * mapping_results.getSumOfSuccesses() / total_domains ) + "%" );
2177                                 }
2178                                 else {
2179                                     System.out.println( ", mapped ratio = n/a (total domains = 0 )" );
2180                                 }
2181                             }
2182                         }
2183                     }
2184                 }
2185                 i++;
2186             } // for( final Phylogeny intree : intrees ) {
2187         }
2188         if ( plus_minus_analysis_high_copy_base_species.size() > 0 ) {
2189             executePlusMinusAnalysis( output_file,
2190                                       plus_minus_analysis_high_copy_base_species,
2191                                       plus_minus_analysis_high_copy_target_species,
2192                                       plus_minus_analysis_high_low_copy_species,
2193                                       gwcd_list,
2194                                       protein_lists_per_species,
2195                                       domain_id_to_go_ids_map,
2196                                       go_id_to_term_map,
2197                                       plus_minus_analysis_numbers );
2198         }
2199         if ( output_protein_lists_for_all_domains ) {
2200             writeProteinListsForAllSpecies( out_dir, protein_lists_per_species, gwcd_list );
2201         }
2202         //        if ( ( intrees != null ) && ( intrees.length > 0 ) && ( inferred_trees != null ) && ( inferred_trees.size() > 0 ) ) {
2203         //            final StringBuilder parameters_sb = createParametersAsString( ignore_dufs,
2204         //                                                                          e_value_max,
2205         //                                                                          max_allowed_overlap,
2206         //                                                                          no_engulfing_overlaps,
2207         //                                                                          cutoff_scores_file );
2208         //            String s = "_";
2209         //            if ( radomize_fitch_parsimony ) {
2210         //                s += random_number_seed_for_fitch_parsimony + "_";
2211         //            }
2212         //            int i = 0;
2213         //            for( final Phylogeny inferred_tree : inferred_trees ) {
2214         //                if ( !inferred_tree.isRooted() ) {
2215         //                    intrees[ 0 ].getRoot().getName();
2216         //                    inferred_tree.r
2217         //                }
2218         //                final String outfile_name = ForesterUtil.removeSuffix( inferred_tree.getName() ) + s;
2219         //                final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator
2220         //                        .createInstance( inferred_tree, gwcd_list );
2221         //                SurfacingUtil.executeParsimonyAnalysis( random_number_seed_for_fitch_parsimony,
2222         //                                                        radomize_fitch_parsimony,
2223         //                                                        outfile_name,
2224         //                                                        domain_parsimony,
2225         //                                                        inferred_tree,
2226         //                                                        domain_id_to_go_ids_map,
2227         //                                                        go_id_to_term_map,
2228         //                                                        go_namespace_limit,
2229         //                                                        parameters_sb.toString() );
2230         //                i++;
2231         //            }
2232         //        }
2233         if ( all_bin_domain_combinations_gained_fitch != null ) {
2234             try {
2235                 executeFitchGainsAnalysis( new File( output_file
2236                                                    + surfacing_hmmpfam.OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2237                                            all_bin_domain_combinations_gained_fitch,
2238                                            all_domains_encountered.size(),
2239                                            all_bin_domain_combinations_encountered,
2240                                            true );
2241             }
2242             catch ( final IOException e ) {
2243                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2244             }
2245         }
2246         if ( all_bin_domain_combinations_lost_fitch != null ) {
2247             try {
2248                 executeFitchGainsAnalysis( new File( output_file
2249                                                    + surfacing_hmmpfam.OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX ),
2250                                            all_bin_domain_combinations_lost_fitch,
2251                                            all_domains_encountered.size(),
2252                                            all_bin_domain_combinations_encountered,
2253                                            false );
2254             }
2255             catch ( final IOException e ) {
2256                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
2257             }
2258         }
2259         final Runtime rt = java.lang.Runtime.getRuntime();
2260         final long free_memory = rt.freeMemory() / 1000000;
2261         final long total_memory = rt.totalMemory() / 1000000;
2262         System.out.println();
2263         System.out.println( "Time for analysis : " + ( new Date().getTime() - analysis_start_time ) + "ms" );
2264         System.out.println( "Total running time: " + ( new Date().getTime() - start_time ) + "ms " );
2265         System.out.println( "Free memory       : " + free_memory + "MB, total memory: " + total_memory + "MB" );
2266         System.out.println();
2267         System.out.println( "If this application is useful to you, please cite:" );
2268         System.out.println( surfacing_hmmpfam.WWW );
2269         System.out.println();
2270         ForesterUtil.programMessage( PRG_NAME, "OK" );
2271         System.out.println();
2272     }
2273
2274     private static void preparePhylogenyForParsimonyAnalyses( final Phylogeny intree,
2275                                                               final String[][] input_file_properties ) {
2276         final String[] genomes = new String[ input_file_properties.length ];
2277         for( int i = 0; i < input_file_properties.length; ++i ) {
2278             if ( intree.getNodes( input_file_properties[ i ][ 1 ] ).size() > 1 ) {
2279                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "node named [" + input_file_properties[ i ][ 1 ]
2280                         + "] is not unique in input tree " + intree.getName() );
2281             }
2282             genomes[ i ] = input_file_properties[ i ][ 1 ];
2283         }
2284         PhylogenyMethods.deleteExternalNodesPositiveSelection( genomes, intree );
2285         for( int i = 0; i < input_file_properties.length; ++i ) {
2286             try {
2287                 intree.getNode( input_file_properties[ i ][ 1 ] );
2288             }
2289             catch ( final IllegalArgumentException e ) {
2290                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "node named [" + input_file_properties[ i ][ 1 ]
2291                         + "] not present/not unique in input tree" );
2292             }
2293         }
2294     }
2295
2296     // public static StringBuffer stringCombinableDomainsMapToStringBuffer(
2297     // final SortedMap<String, CombinableDomains> map ) {
2298     // final StringBuffer sb = new StringBuffer();
2299     // for( final Iterator<String> iter = map.keySet().iterator();
2300     // iter.hasNext(); ) {
2301     // final Object key = iter.next();
2302     // sb.append( ForesterUtil.pad( new StringBuffer( key.toString() ), 18, ' ',
2303     // false ) );
2304     // final CombinableDomains domain_combination = map.get( key );
2305     // sb.append( ForesterUtil.pad( new StringBuffer( "" +
2306     // domain_combination.getNumberOfCombiningDomains() ), 8,
2307     // ' ', false ) );
2308     // sb.append( domain_combination.toStringBuffer() );
2309     // sb.append( ForesterUtil.getLineSeparator() );
2310     // }
2311     // return sb;
2312     // }
2313     private static void printHelp() {
2314         System.out.println();
2315         System.out.println( "Usage:" );
2316         System.out.println();
2317         System.out.println( "% java -Xms256m -Xmx512m -cp forester.jar org.forester.applications."
2318                 + surfacing_hmmpfam.PRG_NAME
2319                 + " [options] <phylogen(y|ies) infile> [external node name 1] [name 2] ... [name n]" );
2320         System.out.println();
2321         System.out.println( " Note: This software might need a significant amount of memory (heap space);" );
2322         System.out
2323                 .println( "       hence use \"-Xms128m -Xmx512m\" (or more) to prevent a \"java.lang.OutOfMemoryError\"." );
2324         System.out.println();
2325         System.out.println( " Options: " );
2326         System.out.println( surfacing_hmmpfam.DETAILEDNESS_OPTION
2327                 + ": level of detail for similarities output file (default:" + DETAILEDNESS_DEFAULT + ")" );
2328         System.out.println( surfacing_hmmpfam.IGNORE_COMBINATION_WITH_SAME_OPTION
2329                 + ": to ignore combinations with self (default: not to ignore)" );
2330         System.out
2331                 .println( surfacing_hmmpfam.IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION
2332                         + ": to ignore domains without combinations in any species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2333         System.out
2334                 .println( surfacing_hmmpfam.IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION
2335                         + ": to ignore domains specific to one species (for similarity calc purposes, not for parsimony analyses) (default: not to ignore)" );
2336         System.out.println( surfacing_hmmpfam.NOT_IGNORE_DUFS_OPTION
2337                 + ": to _not_ ignore DUFs (domains with unknown function) (default: ignore DUFs)" );
2338         System.out
2339                 .println( surfacing_hmmpfam.IGNORE_VIRAL_IDS
2340                         + ": to ignore domains with ids containing 'vir', 'retro', 'transpos', 'phage', or starting with 'rv' or 'gag_'" );
2341         System.out.println( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_OPTION + ": sorting for similarities (default: "
2342                 + DOMAIN_SORT_FILD_DEFAULT + ")" );
2343         System.out.println( surfacing_hmmpfam.OUTPUT_FILE_OPTION + ": name for (main) output file (mandatory)" );
2344         System.out.println( surfacing_hmmpfam.MAX_E_VALUE_OPTION + ": max (inclusive) E-value" );
2345         System.out.println( surfacing_hmmpfam.MAX_ALLOWED_OVERLAP_OPTION + ": maximal allowed domain overlap" );
2346         System.out.println( surfacing_hmmpfam.NO_ENGULFING_OVERLAP_OPTION
2347                 + ": to ignore engulfed lower confidence domains" );
2348         System.out.println( surfacing_hmmpfam.SPECIES_MATRIX_OPTION + ": species matrix" );
2349         System.out.println( surfacing_hmmpfam.SCORING_OPTION + ": scoring (default:" + SCORING_DEFAULT + ")" );
2350         System.out.println( surfacing_hmmpfam.DOMAIN_COUNT_SORT_OPTION + ": sorting for domain counts (default:"
2351                 + DOMAINS_SORT_ORDER_DEFAULT + ")" );
2352         System.out.println( surfacing_hmmpfam.DOMAIN_SIMILARITY_PRINT_OPTION
2353                 + ": domain similarity print option (default:" + DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT + ")" );
2354         System.out.println( surfacing_hmmpfam.CUTOFF_SCORE_FILE_OPTION + ": cutoff score file" );
2355         System.out.println( surfacing_hmmpfam.DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION
2356                 + ": sort by species count first" );
2357         System.out.println( surfacing_hmmpfam.OUTPUT_DIR_OPTION + ": output directory" );
2358         System.out.println( surfacing_hmmpfam.PFAM_TO_GO_FILE_USE_OPTION + ": Pfam to GO mapping file" );
2359         System.out.println( surfacing_hmmpfam.GO_OBO_FILE_USE_OPTION + ": GO terms file (OBO format)" );
2360         System.out.println( surfacing_hmmpfam.GO_NAMESPACE_LIMIT_OPTION + ": limit GO term to one GO namespace" );
2361         System.out.println( surfacing_hmmpfam.PAIRWISE_DOMAIN_COMPARISONS_OPTION
2362                 + "[=<suffix for pairwise comparison output files>]: to perform pairwise comparison based analyses" );
2363         System.out.println( surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION
2364                 + ": species tree, to perform (Dollo, Fitch) parismony analyses" );
2365         System.out.println( surfacing_hmmpfam.DISPLAY_M_HISTOGRAMS_OPTION
2366                 + ": to display multiple histograms (using fluorite)" );
2367         System.out
2368                 .println( JACKNIFE_OPTION
2369                         + ": perform jacknife resampling for domain and binary domain combination based distance matrices [default resamplings: "
2370                         + JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT + "]" );
2371         System.out.println( JACKNIFE_RATIO_OPTION + ": ratio for jacknife resampling [default: "
2372                 + JACKNIFE_RATIO_DEFAULT + "]" );
2373         System.out.println( JACKNIFE_RANDOM_SEED_OPTION
2374                 + ": seed for random number generator for jacknife resampling [default: "
2375                 + JACKNIFE_RANDOM_SEED_DEFAULT + "]" );
2376         //        System.out.println( surfacing.INFER_SPECIES_TREES_OPTION
2377         //                + ": to infer NJ species trees based on shared domains/binary domain combinations" );
2378         System.out
2379                 .println( surfacing_hmmpfam.INPUT_SPECIES_TREE_OPTION
2380                         + "=<treefiles in phyloXML format, separated by #>: to infer domain/binary domain combination gains/losses on given species trees" );
2381         System.out.println( surfacing_hmmpfam.FILTER_POSITIVE_OPTION
2382                 + "=<file>: to filter out proteins not containing at least one domain listed in <file>" );
2383         System.out.println( surfacing_hmmpfam.FILTER_NEGATIVE_OPTION
2384                 + "=<file>: to filter out proteins containing at least one domain listed in <file>" );
2385         System.out.println( surfacing_hmmpfam.FILTER_NEGATIVE_DOMAINS_OPTION
2386                 + "=<file>: to filter out (ignore) domains listed in <file>" );
2387         System.out
2388                 .println( surfacing_hmmpfam.INPUT_FILES_FROM_FILE_OPTION + "=<file>: to read input files from <file>" );
2389         System.out
2390                 .println( surfacing_hmmpfam.RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION
2391                         + "=<seed>: seed for random number generator for Fitch Parsimony analysis (type: long, default: no randomization - given a choice, prefer absence" );
2392         System.out.println( surfacing_hmmpfam.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS
2393                 + ": to consider directedness in binary combinations: e.g. A-B != B-A" );
2394         System.out.println( surfacing_hmmpfam.CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY
2395                 + ": to consider directedness and adjacency in binary combinations" );
2396         System.out
2397                 .println( surfacing_hmmpfam.SEQ_EXTRACT_OPTION
2398                         + "=<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')" );
2399         System.out.println( surfacing_hmmpfam.SECONDARY_FEATURES_PARSIMONY_MAP_FILE
2400                 + "=<file>: to perfom parsimony analysis on secondary features" );
2401         System.out.println( surfacing_hmmpfam.PLUS_MINUS_ANALYSIS_OPTION
2402                 + "=<file>: to presence/absence genome analysis" );
2403         System.out.println( surfacing_hmmpfam.DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS
2404                 + ": to output binary domain combinations for (downstream) graph analysis" );
2405         System.out.println( surfacing_hmmpfam.OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS
2406                 + ": to output all proteins per domain" );
2407         System.out.println();
2408         System.out.println();
2409         System.out.println( "Example: java -Xms128m -Xmx512m -cp path/to/forester.jar"
2410                 + "org.forester.application.surfacing -detail=punctilious -o=TEST.html -pwc=TEST"
2411                 + " -cos=Pfam_ls_22_TC2 -p2g=pfam2go -obo=gene_ontology_edit.obo "
2412                 + "-dc_sort=dom -ignore_with_self -no_singles -e=0.001 -mo=1 -no_eo "
2413                 + "-ds_output=detailed_html -scoring=domains -sort=alpha -" + JACKNIFE_OPTION
2414                 + "=50 human mouse brafl strpu" );
2415         System.out.println();
2416     }
2417
2418     private static void processFilter( final File filter_file, final SortedSet<DomainId> filter ) {
2419         SortedSet<String> filter_str = null;
2420         try {
2421             filter_str = ForesterUtil.file2set( filter_file );
2422         }
2423         catch ( final IOException e ) {
2424             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage() );
2425         }
2426         if ( filter_str != null ) {
2427             for( final String string : filter_str ) {
2428                 filter.add( new DomainId( string ) );
2429             }
2430         }
2431         if ( VERBOSE ) {
2432             System.out.println( "Filter:" );
2433             for( final DomainId domainId : filter ) {
2434                 System.out.println( domainId.getId() );
2435             }
2436         }
2437     }
2438
2439     private static String[][] processInputFileNames( final String[] names ) {
2440         final String[][] input_file_properties = new String[ names.length ][];
2441         for( int i = 0; i < names.length; ++i ) {
2442             if ( names[ i ].indexOf( SEPARATOR_FOR_INPUT_VALUES ) < 0 ) {
2443                 input_file_properties[ i ] = new String[ 3 ];
2444                 input_file_properties[ i ][ 0 ] = names[ i ];
2445                 input_file_properties[ i ][ 1 ] = names[ i ];
2446                 input_file_properties[ i ][ 2 ] = DEFAULT_SEARCH_PARAMETER;
2447             }
2448             else {
2449                 input_file_properties[ i ] = names[ i ].split( surfacing_hmmpfam.SEPARATOR_FOR_INPUT_VALUES + "" );
2450                 if ( input_file_properties[ i ].length != 3 ) {
2451                     ForesterUtil
2452                             .fatalError( surfacing_hmmpfam.PRG_NAME,
2453                                          "properties for the input files (hmmpfam output) are expected "
2454                                                  + "to be in the following format \"<hmmpfam output file>#<species>#<search"
2455                                                  + " parameter>\" (or just one word, which is both the filename and the species id), instead received \""
2456                                                  + names[ i ] + "\"" );
2457                 }
2458             }
2459             final String error = ForesterUtil.isReadableFile( new File( input_file_properties[ i ][ 0 ] ) );
2460             if ( !ForesterUtil.isEmpty( error ) ) {
2461                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, error );
2462             }
2463         }
2464         return input_file_properties;
2465     }
2466
2467     private static void processPlusMinusAnalysisOption( final CommandLineArguments cla,
2468                                                         final List<String> high_copy_base,
2469                                                         final List<String> high_copy_target,
2470                                                         final List<String> low_copy,
2471                                                         final List<Object> numbers ) {
2472         if ( cla.isOptionSet( surfacing_hmmpfam.PLUS_MINUS_ANALYSIS_OPTION ) ) {
2473             if ( !cla.isOptionValueSet( surfacing_hmmpfam.PLUS_MINUS_ANALYSIS_OPTION ) ) {
2474                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "no value for 'plus-minus' file: -"
2475                         + surfacing_hmmpfam.PLUS_MINUS_ANALYSIS_OPTION + "=<file>" );
2476             }
2477             final File plus_minus_file = new File( cla.getOptionValue( surfacing_hmmpfam.PLUS_MINUS_ANALYSIS_OPTION ) );
2478             final String msg = ForesterUtil.isReadableFile( plus_minus_file );
2479             if ( !ForesterUtil.isEmpty( msg ) ) {
2480                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "can not read from \"" + plus_minus_file + "\": "
2481                         + msg );
2482             }
2483             processPlusMinusFile( plus_minus_file, high_copy_base, high_copy_target, low_copy, numbers );
2484         }
2485     }
2486
2487     // First numbers is minimal difference, second is factor.
2488     private static void processPlusMinusFile( final File plus_minus_file,
2489                                               final List<String> high_copy_base,
2490                                               final List<String> high_copy_target,
2491                                               final List<String> low_copy,
2492                                               final List<Object> numbers ) {
2493         Set<String> species_set = null;
2494         int min_diff = PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT;
2495         double factor = PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT;
2496         try {
2497             species_set = ForesterUtil.file2set( plus_minus_file );
2498         }
2499         catch ( final IOException e ) {
2500             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getMessage() );
2501         }
2502         if ( species_set != null ) {
2503             for( final String species : species_set ) {
2504                 final String species_trimmed = species.substring( 1 );
2505                 if ( species.startsWith( "+" ) ) {
2506                     if ( low_copy.contains( species_trimmed ) ) {
2507                         ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
2508                                                  "species/genome names can not appear with both '+' and '-' suffix, as appears the case for: \""
2509                                                          + species_trimmed + "\"" );
2510                     }
2511                     high_copy_base.add( species_trimmed );
2512                 }
2513                 else if ( species.startsWith( "*" ) ) {
2514                     if ( low_copy.contains( species_trimmed ) ) {
2515                         ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
2516                                                  "species/genome names can not appear with both '*' and '-' suffix, as appears the case for: \""
2517                                                          + species_trimmed + "\"" );
2518                     }
2519                     high_copy_target.add( species_trimmed );
2520                 }
2521                 else if ( species.startsWith( "-" ) ) {
2522                     if ( high_copy_base.contains( species_trimmed ) || high_copy_target.contains( species_trimmed ) ) {
2523                         ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
2524                                                  "species/genome names can not appear with both '+' or '*' and '-' suffix, as appears the case for: \""
2525                                                          + species_trimmed + "\"" );
2526                     }
2527                     low_copy.add( species_trimmed );
2528                 }
2529                 else if ( species.startsWith( "$D" ) ) {
2530                     try {
2531                         min_diff = Integer.parseInt( species.substring( 3 ) );
2532                     }
2533                     catch ( final NumberFormatException e ) {
2534                         ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
2535                                                  "could not parse integer value for minimal difference from: \""
2536                                                          + species.substring( 3 ) + "\"" );
2537                     }
2538                 }
2539                 else if ( species.startsWith( "$F" ) ) {
2540                     try {
2541                         factor = Double.parseDouble( species.substring( 3 ) );
2542                     }
2543                     catch ( final NumberFormatException e ) {
2544                         ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME,
2545                                                  "could not parse double value for factor from: \""
2546                                                          + species.substring( 3 ) + "\"" );
2547                     }
2548                 }
2549                 else if ( species.startsWith( "#" ) ) {
2550                     // Comment, ignore.
2551                 }
2552                 else {
2553                     ForesterUtil
2554                             .fatalError( surfacing_hmmpfam.PRG_NAME,
2555                                          "species/genome names in 'plus minus' file must begin with '*' (high copy target genome), '+' (high copy base genomes), '-' (low copy genomes), '$D=<integer>' minimal Difference (default is 1), '$F=<double>' factor (default is 1.0), double), or '#' (ignore) suffix, encountered: \""
2556                                                  + species + "\"" );
2557                 }
2558                 numbers.add( new Integer( min_diff + "" ) );
2559                 numbers.add( new Double( factor + "" ) );
2560             }
2561         }
2562         else {
2563             ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, "'plus minus' file [" + plus_minus_file
2564                     + "] appears empty" );
2565         }
2566     }
2567
2568     private static void writeProteinListsForAllSpecies( final File output_dir,
2569                                                         final SortedMap<Species, List<Protein>> protein_lists_per_species,
2570                                                         final List<GenomeWideCombinableDomains> gwcd_list ) {
2571         final SortedSet<DomainId> all_domains = new TreeSet<DomainId>();
2572         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
2573             all_domains.addAll( gwcd.getAllDomainIds() );
2574         }
2575         for( final DomainId domain : all_domains ) {
2576             final File out = new File( output_dir + ForesterUtil.FILE_SEPARATOR + domain + SEQ_EXTRACT_SUFFIX );
2577             SurfacingUtil.checkForOutputFileWriteability( out );
2578             try {
2579                 final Writer proteins_file_writer = new BufferedWriter( new FileWriter( out ) );
2580                 SurfacingUtil.extractProteinNames( protein_lists_per_species, domain, proteins_file_writer, "\t", null );
2581                 proteins_file_writer.close();
2582             }
2583             catch ( final IOException e ) {
2584                 ForesterUtil.fatalError( surfacing_hmmpfam.PRG_NAME, e.getLocalizedMessage() );
2585             }
2586             ForesterUtil.programMessage( surfacing_hmmpfam.PRG_NAME, "Wrote proteins list to \"" + out + "\"" );
2587         }
2588     }
2589 }