forester/java/src/org/forester/surfacing/SurfacingUtil.java

   1 // $Id:
   2 //
   3 // FORESTER -- software libraries and applications
   4 // for evolutionary biology research and applications.
   5 //
   6 // Copyright (C) 2008-2009 Christian M. Zmasek
   7 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   8 // All rights reserved
   9 //
  10 // This library is free software; you can redistribute it and/or
  11 // modify it under the terms of the GNU Lesser General Public
  12 // License as published by the Free Software Foundation; either
  13 // version 2.1 of the License, or (at your option) any later version.
  14 //
  15 // This library is distributed in the hope that it will be useful,
  16 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18 // Lesser General Public License for more details.
  19 //
  20 // You should have received a copy of the GNU Lesser General Public
  21 // License along with this library; if not, write to the Free Software
  22 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  23 //
  24 // Contact: phylosoft @ gmail . com
  25 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
  26
  27 package org.forester.surfacing;
  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.text.DecimalFormat;
  35 import java.text.NumberFormat;
  36 import java.util.ArrayList;
  37 import java.util.Arrays;
  38 import java.util.Collections;
  39 import java.util.Comparator;
  40 import java.util.HashMap;
  41 import java.util.HashSet;
  42 import java.util.Iterator;
  43 import java.util.List;
  44 import java.util.Map;
  45 import java.util.Map.Entry;
  46 import java.util.PriorityQueue;
  47 import java.util.Set;
  48 import java.util.SortedMap;
  49 import java.util.SortedSet;
  50 import java.util.TreeMap;
  51 import java.util.TreeSet;
  52 import java.util.regex.Matcher;
  53 import java.util.regex.Pattern;
  54
  55 import org.forester.application.surfacing;
  56 import org.forester.evoinference.distance.NeighborJoining;
  57 import org.forester.evoinference.matrix.character.BasicCharacterStateMatrix;
  58 import org.forester.evoinference.matrix.character.CharacterStateMatrix;
  59 import org.forester.evoinference.matrix.character.CharacterStateMatrix.BinaryStates;
  60 import org.forester.evoinference.matrix.character.CharacterStateMatrix.Format;
  61 import org.forester.evoinference.matrix.character.CharacterStateMatrix.GainLossStates;
  62 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
  63 import org.forester.evoinference.matrix.distance.DistanceMatrix;
  64 import org.forester.go.GoId;
  65 import org.forester.go.GoNameSpace;
  66 import org.forester.go.GoTerm;
  67 import org.forester.go.PfamToGoMapping;
  68 import org.forester.io.parsers.nexus.NexusConstants;
  69 import org.forester.io.writers.PhylogenyWriter;
  70 import org.forester.phylogeny.Phylogeny;
  71 import org.forester.phylogeny.PhylogenyMethods;
  72 import org.forester.phylogeny.PhylogenyNode;
  73 import org.forester.phylogeny.PhylogenyNode.NH_CONVERSION_SUPPORT_VALUE_STYLE;
  74 import org.forester.phylogeny.data.BinaryCharacters;
  75 import org.forester.phylogeny.data.Confidence;
  76 import org.forester.phylogeny.data.Taxonomy;
  77 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  78 import org.forester.protein.BasicDomain;
  79 import org.forester.protein.BasicProtein;
  80 import org.forester.protein.BinaryDomainCombination;
  81 import org.forester.protein.Domain;
  82 import org.forester.protein.DomainId;
  83 import org.forester.protein.Protein;
  84 import org.forester.species.Species;
  85 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  86 import org.forester.surfacing.DomainSimilarityCalculator.GoAnnotationOutput;
  87 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  88 import org.forester.util.AsciiHistogram;
  89 import org.forester.util.BasicDescriptiveStatistics;
  90 import org.forester.util.BasicTable;
  91 import org.forester.util.BasicTableParser;
  92 import org.forester.util.DescriptiveStatistics;
  93 import org.forester.util.ForesterUtil;
  94
  95 public final class SurfacingUtil {
  96
  97     private final static NumberFormat       FORMATTER_3                      = new DecimalFormat( "0.000" );
  98     private static final Comparator<Domain> ASCENDING_CONFIDENCE_VALUE_ORDER = new Comparator<Domain>() {
  99
 100                                                                                  @Override
 101                                                                                  public int compare( final Domain d1,
 102                                                                                                      final Domain d2 ) {
 103                                                                                      if ( d1.getPerSequenceEvalue() < d2
 104                                                                                              .getPerSequenceEvalue() ) {
 105                                                                                          return -1;
 106                                                                                      }
 107                                                                                      else if ( d1
 108                                                                                              .getPerSequenceEvalue() > d2
 109                                                                                              .getPerSequenceEvalue() ) {
 110                                                                                          return 1;
 111                                                                                      }
 112                                                                                      else {
 113                                                                                          return d1.compareTo( d2 );
 114                                                                                      }
 115                                                                                  }
 116                                                                              };
 117     public final static Pattern             PATTERN_SP_STYLE_TAXONOMY        = Pattern.compile( "^[A-Z0-9]{3,5}$" );
 118     private static final boolean            USE_LAST                         = true;
 119
 120     private SurfacingUtil() {
 121         // Hidden constructor.
 122     }
 123
 124     public static void addAllBinaryDomainCombinationToSet( final GenomeWideCombinableDomains genome,
 125                                                            final SortedSet<BinaryDomainCombination> binary_domain_combinations ) {
 126         final SortedMap<DomainId, CombinableDomains> all_cd = genome.getAllCombinableDomainsIds();
 127         for( final DomainId domain_id : all_cd.keySet() ) {
 128             binary_domain_combinations.addAll( all_cd.get( domain_id ).toBinaryDomainCombinations() );
 129         }
 130     }
 131
 132     public static void addAllDomainIdsToSet( final GenomeWideCombinableDomains genome,
 133                                              final SortedSet<DomainId> domain_ids ) {
 134         final SortedSet<DomainId> domains = genome.getAllDomainIds();
 135         for( final DomainId domain : domains ) {
 136             domain_ids.add( domain );
 137         }
 138     }
 139
 140     public static void addHtmlHead( final Writer w, final String title ) throws IOException {
 141         w.write( SurfacingConstants.NL );
 142         w.write( "<head>" );
 143         w.write( "<title>" );
 144         w.write( title );
 145         w.write( "</title>" );
 146         w.write( SurfacingConstants.NL );
 147         w.write( "<style>" );
 148         w.write( SurfacingConstants.NL );
 149         w.write( "a:visited { color : #6633FF; text-decoration : none; }" );
 150         w.write( SurfacingConstants.NL );
 151         w.write( "a:link { color : #6633FF; text-decoration : none; }" );
 152         w.write( SurfacingConstants.NL );
 153         w.write( "a:active { color : #99FF00; text-decoration : none; }" );
 154         w.write( SurfacingConstants.NL );
 155         w.write( "a:hover { color : #FFFFFF; background-color : #99FF00; text-decoration : none; }" );
 156         w.write( SurfacingConstants.NL );
 157         w.write( "td { text-align: left; vertical-align: top; font-family: Verdana, Arial, Helvetica; font-size: 8pt}" );
 158         w.write( SurfacingConstants.NL );
 159         w.write( "h1 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 18pt; font-weight: bold }" );
 160         w.write( SurfacingConstants.NL );
 161         w.write( "h2 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 16pt; font-weight: bold }" );
 162         w.write( SurfacingConstants.NL );
 163         w.write( "</style>" );
 164         w.write( SurfacingConstants.NL );
 165         w.write( "</head>" );
 166         w.write( SurfacingConstants.NL );
 167     }
 168
 169     public static DescriptiveStatistics calculateDescriptiveStatisticsForMeanValues( final Set<DomainSimilarity> similarities ) {
 170         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 171         for( final DomainSimilarity similarity : similarities ) {
 172             stats.addValue( similarity.getMeanSimilarityScore() );
 173         }
 174         return stats;
 175     }
 176
 177     public static int calculateOverlap( final Domain domain, final List<Boolean> covered_positions ) {
 178         int overlap_count = 0;
 179         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
 180             if ( ( i < covered_positions.size() ) && ( covered_positions.get( i ) == true ) ) {
 181                 ++overlap_count;
 182             }
 183         }
 184         return overlap_count;
 185     }
 186
 187     public static void checkForOutputFileWriteability( final File outfile ) {
 188         final String error = ForesterUtil.isWritableFile( outfile );
 189         if ( !ForesterUtil.isEmpty( error ) ) {
 190             ForesterUtil.fatalError( surfacing.PRG_NAME, error );
 191         }
 192     }
 193
 194     public static void collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
 195                                                                                            final BinaryDomainCombination.DomainCombinationType dc_type,
 196                                                                                            final List<BinaryDomainCombination> all_binary_domains_combination_gained,
 197                                                                                            final boolean get_gains ) {
 198         final SortedSet<String> sorted_ids = new TreeSet<String>();
 199         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
 200             sorted_ids.add( matrix.getIdentifier( i ) );
 201         }
 202         for( final String id : sorted_ids ) {
 203             for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
 204                 if ( ( get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) )
 205                         || ( !get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.LOSS ) ) ) {
 206                     if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) {
 207                         all_binary_domains_combination_gained.add( AdjactantDirectedBinaryDomainCombination
 208                                 .createInstance( matrix.getCharacter( c ) ) );
 209                     }
 210                     else if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED ) {
 211                         all_binary_domains_combination_gained.add( DirectedBinaryDomainCombination
 212                                 .createInstance( matrix.getCharacter( c ) ) );
 213                     }
 214                     else {
 215                         all_binary_domains_combination_gained.add( BasicBinaryDomainCombination.createInstance( matrix
 216                                 .getCharacter( c ) ) );
 217                     }
 218                 }
 219             }
 220         }
 221     }
 222
 223     public static Map<DomainId, List<GoId>> createDomainIdToGoIdMap( final List<PfamToGoMapping> pfam_to_go_mappings ) {
 224         final Map<DomainId, List<GoId>> domain_id_to_go_ids_map = new HashMap<DomainId, List<GoId>>( pfam_to_go_mappings
 225                 .size() );
 226         for( final PfamToGoMapping pfam_to_go : pfam_to_go_mappings ) {
 227             if ( !domain_id_to_go_ids_map.containsKey( pfam_to_go.getKey() ) ) {
 228                 domain_id_to_go_ids_map.put( pfam_to_go.getKey(), new ArrayList<GoId>() );
 229             }
 230             domain_id_to_go_ids_map.get( pfam_to_go.getKey() ).add( pfam_to_go.getValue() );
 231         }
 232         return domain_id_to_go_ids_map;
 233     }
 234
 235     public static Map<DomainId, Set<String>> createDomainIdToSecondaryFeaturesMap( final File secondary_features_map_file )
 236             throws IOException {
 237         final BasicTable<String> primary_table = BasicTableParser.parse( secondary_features_map_file, '\t' );
 238         final Map<DomainId, Set<String>> map = new TreeMap<DomainId, Set<String>>();
 239         for( int r = 0; r < primary_table.getNumberOfRows(); ++r ) {
 240             final DomainId domain_id = new DomainId( primary_table.getValue( 0, r ) );
 241             if ( !map.containsKey( domain_id ) ) {
 242                 map.put( domain_id, new HashSet<String>() );
 243             }
 244             map.get( domain_id ).add( primary_table.getValue( 1, r ) );
 245         }
 246         return map;
 247     }
 248
 249     public static Phylogeny createNjTreeBasedOnMatrixToFile( final File nj_tree_outfile, final DistanceMatrix distance ) {
 250         checkForOutputFileWriteability( nj_tree_outfile );
 251         final NeighborJoining nj = NeighborJoining.createInstance();
 252         final Phylogeny phylogeny = nj.execute( ( BasicSymmetricalDistanceMatrix ) distance );
 253         phylogeny.setName( nj_tree_outfile.getName() );
 254         writePhylogenyToFile( phylogeny, nj_tree_outfile.toString() );
 255         return phylogeny;
 256     }
 257
 258     public static Map<String, Integer> createTaxCodeToIdMap( final Phylogeny phy ) {
 259         final Map<String, Integer> m = new HashMap<String, Integer>();
 260         for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
 261             final PhylogenyNode n = iter.next();
 262             if ( n.getNodeData().isHasTaxonomy() ) {
 263                 final Taxonomy t = n.getNodeData().getTaxonomy();
 264                 final String c = t.getTaxonomyCode();
 265                 if ( !ForesterUtil.isEmpty( c ) ) {
 266                     if ( n.getNodeData().getTaxonomy() == null ) {
 267                         ForesterUtil.fatalError( surfacing.PRG_NAME, "no taxonomy id for node " + n );
 268                     }
 269                     final String id = n.getNodeData().getTaxonomy().getIdentifier().getValue();
 270                     if ( ForesterUtil.isEmpty( id ) ) {
 271                         ForesterUtil.fatalError( surfacing.PRG_NAME, "no taxonomy id for node " + n );
 272                     }
 273                     if ( m.containsKey( c ) ) {
 274                         ForesterUtil.fatalError( surfacing.PRG_NAME, "taxonomy code " + c + " is not unique" );
 275                     }
 276                     final int iid = Integer.valueOf( id );
 277                     if ( m.containsValue( iid ) ) {
 278                         ForesterUtil.fatalError( surfacing.PRG_NAME, "taxonomy id " + iid + " is not unique" );
 279                     }
 280                     m.put( c, iid );
 281                 }
 282             }
 283             else {
 284                 ForesterUtil.fatalError( surfacing.PRG_NAME, "no taxonomy for node " + n );
 285             }
 286         }
 287         return m;
 288     }
 289
 290     public static void decoratePrintableDomainSimilarities( final SortedSet<DomainSimilarity> domain_similarities,
 291                                                             final Detailedness detailedness,
 292                                                             final GoAnnotationOutput go_annotation_output,
 293                                                             final Map<GoId, GoTerm> go_id_to_term_map,
 294                                                             final GoNameSpace go_namespace_limit ) {
 295         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) ) {
 296             throw new IllegalArgumentException( "attempt to use a GO namespace limit without a GO id to term map" );
 297         }
 298         for( final DomainSimilarity domain_similarity : domain_similarities ) {
 299             if ( domain_similarity instanceof PrintableDomainSimilarity ) {
 300                 final PrintableDomainSimilarity printable_domain_similarity = ( PrintableDomainSimilarity ) domain_similarity;
 301                 printable_domain_similarity.setDetailedness( detailedness );
 302                 printable_domain_similarity.setGoAnnotationOutput( go_annotation_output );
 303                 printable_domain_similarity.setGoIdToTermMap( go_id_to_term_map );
 304                 printable_domain_similarity.setGoNamespaceLimit( go_namespace_limit );
 305             }
 306         }
 307     }
 308
 309     public static void doit( final List<Protein> proteins,
 310                              final List<DomainId> query_domain_ids_nc_order,
 311                              final Writer out,
 312                              final String separator,
 313                              final String limit_to_species,
 314                              final Map<String, List<Integer>> average_protein_lengths_by_dc ) throws IOException {
 315         for( final Protein protein : proteins ) {
 316             if ( ForesterUtil.isEmpty( limit_to_species )
 317                     || protein.getSpecies().getSpeciesId().equalsIgnoreCase( limit_to_species ) ) {
 318                 if ( protein.contains( query_domain_ids_nc_order, true ) ) {
 319                     out.write( protein.getSpecies().getSpeciesId() );
 320                     out.write( separator );
 321                     out.write( protein.getProteinId().getId() );
 322                     out.write( separator );
 323                     out.write( "[" );
 324                     final Set<DomainId> visited_domain_ids = new HashSet<DomainId>();
 325                     boolean first = true;
 326                     for( final Domain domain : protein.getProteinDomains() ) {
 327                         if ( !visited_domain_ids.contains( domain.getDomainId() ) ) {
 328                             visited_domain_ids.add( domain.getDomainId() );
 329                             if ( first ) {
 330                                 first = false;
 331                             }
 332                             else {
 333                                 out.write( " " );
 334                             }
 335                             out.write( domain.getDomainId().getId() );
 336                             out.write( " {" );
 337                             out.write( "" + domain.getTotalCount() );
 338                             out.write( "}" );
 339                         }
 340                     }
 341                     out.write( "]" );
 342                     out.write( separator );
 343                     if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 344                             .equals( SurfacingConstants.NONE ) ) ) {
 345                         out.write( protein.getDescription() );
 346                     }
 347                     out.write( separator );
 348                     if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 349                             .equals( SurfacingConstants.NONE ) ) ) {
 350                         out.write( protein.getAccession() );
 351                     }
 352                     out.write( SurfacingConstants.NL );
 353                 }
 354             }
 355         }
 356         out.flush();
 357     }
 358
 359     public static void domainsPerProteinsStatistics( final String genome,
 360                                                      final List<Protein> protein_list,
 361                                                      final DescriptiveStatistics all_genomes_domains_per_potein_stats,
 362                                                      final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
 363                                                      final SortedSet<String> domains_which_are_always_single,
 364                                                      final SortedSet<String> domains_which_are_sometimes_single_sometimes_not,
 365                                                      final SortedSet<String> domains_which_never_single,
 366                                                      final Writer writer ) {
 367         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 368         for( final Protein protein : protein_list ) {
 369             final int domains = protein.getNumberOfProteinDomains();
 370             //System.out.println( domains );
 371             stats.addValue( domains );
 372             all_genomes_domains_per_potein_stats.addValue( domains );
 373             if ( !all_genomes_domains_per_potein_histo.containsKey( domains ) ) {
 374                 all_genomes_domains_per_potein_histo.put( domains, 1 );
 375             }
 376             else {
 377                 all_genomes_domains_per_potein_histo.put( domains,
 378                                                           1 + all_genomes_domains_per_potein_histo.get( domains ) );
 379             }
 380             if ( domains == 1 ) {
 381                 final String domain = protein.getProteinDomain( 0 ).getDomainId().getId();
 382                 if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
 383                     if ( domains_which_never_single.contains( domain ) ) {
 384                         domains_which_never_single.remove( domain );
 385                         domains_which_are_sometimes_single_sometimes_not.add( domain );
 386                     }
 387                     else {
 388                         domains_which_are_always_single.add( domain );
 389                     }
 390                 }
 391             }
 392             else if ( domains > 1 ) {
 393                 for( final Domain d : protein.getProteinDomains() ) {
 394                     final String domain = d.getDomainId().getId();
 395                     // System.out.println( domain );
 396                     if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
 397                         if ( domains_which_are_always_single.contains( domain ) ) {
 398                             domains_which_are_always_single.remove( domain );
 399                             domains_which_are_sometimes_single_sometimes_not.add( domain );
 400                         }
 401                         else {
 402                             domains_which_never_single.add( domain );
 403                         }
 404                     }
 405                 }
 406             }
 407         }
 408         try {
 409             writer.write( genome );
 410             writer.write( "\t" );
 411             if ( stats.getN() >= 1 ) {
 412                 writer.write( stats.arithmeticMean() + "" );
 413                 writer.write( "\t" );
 414                 if ( stats.getN() >= 2 ) {
 415                     writer.write( stats.sampleStandardDeviation() + "" );
 416                 }
 417                 else {
 418                     writer.write( "" );
 419                 }
 420                 writer.write( "\t" );
 421                 writer.write( stats.median() + "" );
 422                 writer.write( "\t" );
 423                 writer.write( stats.getN() + "" );
 424                 writer.write( "\t" );
 425                 writer.write( stats.getMin() + "" );
 426                 writer.write( "\t" );
 427                 writer.write( stats.getMax() + "" );
 428             }
 429             else {
 430                 writer.write( "\t" );
 431                 writer.write( "\t" );
 432                 writer.write( "\t" );
 433                 writer.write( "0" );
 434                 writer.write( "\t" );
 435                 writer.write( "\t" );
 436             }
 437             writer.write( "\n" );
 438         }
 439         catch ( final IOException e ) {
 440             e.printStackTrace();
 441         }
 442     }
 443
 444     public static void executeDomainLengthAnalysis( final String[][] input_file_properties,
 445                                                     final int number_of_genomes,
 446                                                     final DomainLengthsTable domain_lengths_table,
 447                                                     final File outfile ) throws IOException {
 448         final DecimalFormat df = new DecimalFormat( "#.00" );
 449         checkForOutputFileWriteability( outfile );
 450         final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
 451         out.write( "MEAN BASED STATISTICS PER SPECIES" );
 452         out.write( ForesterUtil.LINE_SEPARATOR );
 453         out.write( domain_lengths_table.createMeanBasedStatisticsPerSpeciesTable().toString() );
 454         out.write( ForesterUtil.LINE_SEPARATOR );
 455         out.write( ForesterUtil.LINE_SEPARATOR );
 456         final List<DomainLengths> domain_lengths_list = domain_lengths_table.getDomainLengthsList();
 457         out.write( "OUTLIER SPECIES PER DOMAIN (Z>=1.5)" );
 458         out.write( ForesterUtil.LINE_SEPARATOR );
 459         for( final DomainLengths domain_lengths : domain_lengths_list ) {
 460             final List<Species> species_list = domain_lengths.getMeanBasedOutlierSpecies( 1.5 );
 461             if ( species_list.size() > 0 ) {
 462                 out.write( domain_lengths.getDomainId() + "\t" );
 463                 for( final Species species : species_list ) {
 464                     out.write( species + "\t" );
 465                 }
 466                 out.write( ForesterUtil.LINE_SEPARATOR );
 467                 // DescriptiveStatistics stats_for_domain = domain_lengths
 468                 //         .calculateMeanBasedStatistics();
 469                 //AsciiHistogram histo = new AsciiHistogram( stats_for_domain );
 470                 //System.out.println( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 471             }
 472         }
 473         out.write( ForesterUtil.LINE_SEPARATOR );
 474         out.write( ForesterUtil.LINE_SEPARATOR );
 475         out.write( "OUTLIER SPECIES (Z 1.0)" );
 476         out.write( ForesterUtil.LINE_SEPARATOR );
 477         final DescriptiveStatistics stats_for_all_species = domain_lengths_table
 478                 .calculateMeanBasedStatisticsForAllSpecies();
 479         out.write( stats_for_all_species.asSummary() );
 480         out.write( ForesterUtil.LINE_SEPARATOR );
 481         final AsciiHistogram histo = new AsciiHistogram( stats_for_all_species );
 482         out.write( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 483         out.write( ForesterUtil.LINE_SEPARATOR );
 484         final double population_sd = stats_for_all_species.sampleStandardDeviation();
 485         final double population_mean = stats_for_all_species.arithmeticMean();
 486         for( final Species species : domain_lengths_table.getSpecies() ) {
 487             final double x = domain_lengths_table.calculateMeanBasedStatisticsForSpecies( species ).arithmeticMean();
 488             final double z = ( x - population_mean ) / population_sd;
 489             out.write( species + "\t" + z );
 490             out.write( ForesterUtil.LINE_SEPARATOR );
 491         }
 492         out.write( ForesterUtil.LINE_SEPARATOR );
 493         for( final Species species : domain_lengths_table.getSpecies() ) {
 494             final DescriptiveStatistics stats_for_species = domain_lengths_table
 495                     .calculateMeanBasedStatisticsForSpecies( species );
 496             final double x = stats_for_species.arithmeticMean();
 497             final double z = ( x - population_mean ) / population_sd;
 498             if ( ( z <= -1.0 ) || ( z >= 1.0 ) ) {
 499                 out.write( species + "\t" + df.format( z ) + "\t" + stats_for_species.asSummary() );
 500                 out.write( ForesterUtil.LINE_SEPARATOR );
 501             }
 502         }
 503         out.close();
 504         //        final List<HistogramData> histogram_datas = new ArrayList<HistogramData>();
 505         //        for( int i = 0; i < number_of_genomes; ++i ) {
 506         //            final Species species = new BasicSpecies( input_file_properties[ i ][ 0 ] );
 507         //            histogram_datas
 508         //                    .add( new HistogramData( species.toString(), domain_lengths_table
 509         //                            .calculateMeanBasedStatisticsForSpecies( species )
 510         //                            .getDataAsDoubleArray(), 5, 600, null, 60 ) );
 511         //        }
 512         //        final HistogramsFrame hf = new HistogramsFrame( histogram_datas );
 513         //        hf.setVisible( true );
 514         System.gc();
 515     }
 516
 517     /**
 518      *
 519      * @param all_binary_domains_combination_lost_fitch
 520      * @param consider_directedness_and_adjacency_for_bin_combinations
 521      * @param all_binary_domains_combination_gained if null ignored, otherwise this is to list all binary domain combinations
 522      * which were gained under unweighted (Fitch) parsimony.
 523      */
 524     public static void executeParsimonyAnalysis( final long random_number_seed_for_fitch_parsimony,
 525                                                  final boolean radomize_fitch_parsimony,
 526                                                  final String outfile_name,
 527                                                  final DomainParsimonyCalculator domain_parsimony,
 528                                                  final Phylogeny phylogeny,
 529                                                  final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 530                                                  final Map<GoId, GoTerm> go_id_to_term_map,
 531                                                  final GoNameSpace go_namespace_limit,
 532                                                  final String parameters_str,
 533                                                  final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
 534                                                  final SortedSet<DomainId> positive_filter,
 535                                                  final boolean output_binary_domain_combinations_for_graphs,
 536                                                  final List<BinaryDomainCombination> all_binary_domains_combination_gained_fitch,
 537                                                  final List<BinaryDomainCombination> all_binary_domains_combination_lost_fitch,
 538                                                  final BinaryDomainCombination.DomainCombinationType dc_type,
 539                                                  final Map<String, DescriptiveStatistics> protein_length_stats_by_dc,
 540                                                  final Map<String, DescriptiveStatistics> domain_number_stats_by_dc,
 541                                                  final Map<String, DescriptiveStatistics> domain_length_stats_by_domain,
 542                                                  final Map<String, Integer> tax_code_to_id_map,
 543                                                  final boolean write_to_nexus ) {
 544         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 545         final String date_time = ForesterUtil.getCurrentDateTime();
 546         final SortedSet<String> all_pfams_encountered = new TreeSet<String>();
 547         final SortedSet<String> all_pfams_gained_as_domains = new TreeSet<String>();
 548         final SortedSet<String> all_pfams_lost_as_domains = new TreeSet<String>();
 549         final SortedSet<String> all_pfams_gained_as_dom_combinations = new TreeSet<String>();
 550         final SortedSet<String> all_pfams_lost_as_dom_combinations = new TreeSet<String>();
 551         if ( write_to_nexus ) {
 552             writeToNexus( outfile_name, domain_parsimony, phylogeny );
 553         }
 554         // DOLLO DOMAINS
 555         // -------------
 556         Phylogeny local_phylogeny_l = phylogeny.copy();
 557         if ( ( positive_filter != null ) && ( positive_filter.size() > 0 ) ) {
 558             domain_parsimony.executeDolloParsimonyOnDomainPresence( positive_filter );
 559         }
 560         else {
 561             domain_parsimony.executeDolloParsimonyOnDomainPresence();
 562         }
 563         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 564                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 565         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 566                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 567         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 568                                                            CharacterStateMatrix.GainLossStates.GAIN,
 569                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_D,
 570                                                            sep,
 571                                                            ForesterUtil.LINE_SEPARATOR,
 572                                                            null );
 573         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 574                                                            CharacterStateMatrix.GainLossStates.LOSS,
 575                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_D,
 576                                                            sep,
 577                                                            ForesterUtil.LINE_SEPARATOR,
 578                                                            null );
 579         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 580                 + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_D, sep, ForesterUtil.LINE_SEPARATOR, null );
 581         //HTML:
 582         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 583                                        go_id_to_term_map,
 584                                        go_namespace_limit,
 585                                        false,
 586                                        domain_parsimony.getGainLossMatrix(),
 587                                        CharacterStateMatrix.GainLossStates.GAIN,
 588                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D,
 589                                        sep,
 590                                        ForesterUtil.LINE_SEPARATOR,
 591                                        "Dollo Parsimony | Gains | Domains",
 592                                        "+",
 593                                        domain_id_to_secondary_features_maps,
 594                                        all_pfams_encountered,
 595                                        all_pfams_gained_as_domains,
 596                                        "_dollo_gains_d",
 597                                        tax_code_to_id_map );
 598         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 599                                        go_id_to_term_map,
 600                                        go_namespace_limit,
 601                                        false,
 602                                        domain_parsimony.getGainLossMatrix(),
 603                                        CharacterStateMatrix.GainLossStates.LOSS,
 604                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D,
 605                                        sep,
 606                                        ForesterUtil.LINE_SEPARATOR,
 607                                        "Dollo Parsimony | Losses | Domains",
 608                                        "-",
 609                                        domain_id_to_secondary_features_maps,
 610                                        all_pfams_encountered,
 611                                        all_pfams_lost_as_domains,
 612                                        "_dollo_losses_d",
 613                                        tax_code_to_id_map );
 614         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 615                                        go_id_to_term_map,
 616                                        go_namespace_limit,
 617                                        false,
 618                                        domain_parsimony.getGainLossMatrix(),
 619                                        null,
 620                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D,
 621                                        sep,
 622                                        ForesterUtil.LINE_SEPARATOR,
 623                                        "Dollo Parsimony | Present | Domains",
 624                                        "",
 625                                        domain_id_to_secondary_features_maps,
 626                                        all_pfams_encountered,
 627                                        null,
 628                                        "_dollo_present_d",
 629                                        tax_code_to_id_map );
 630         preparePhylogeny( local_phylogeny_l,
 631                           domain_parsimony,
 632                           date_time,
 633                           "Dollo parsimony on domain presence/absence",
 634                           "dollo_on_domains_" + outfile_name,
 635                           parameters_str );
 636         SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 637                 + surfacing.DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 638         try {
 639             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, true, outfile_name, "_dollo_all_gains_d" );
 640             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, false, outfile_name, "_dollo_all_losses_d" );
 641         }
 642         catch ( final IOException e ) {
 643             e.printStackTrace();
 644             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
 645         }
 646         if ( domain_parsimony.calculateNumberOfBinaryDomainCombination() > 0 ) {
 647             // FITCH DOMAIN COMBINATIONS
 648             // -------------------------
 649             local_phylogeny_l = phylogeny.copy();
 650             String randomization = "no";
 651             if ( radomize_fitch_parsimony ) {
 652                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( random_number_seed_for_fitch_parsimony );
 653                 randomization = "yes, seed = " + random_number_seed_for_fitch_parsimony;
 654             }
 655             else {
 656                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( USE_LAST );
 657             }
 658             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 659                     + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 660             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 661                     + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 662             SurfacingUtil
 663                     .writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 664                                                           CharacterStateMatrix.GainLossStates.GAIN,
 665                                                           outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_BC,
 666                                                           sep,
 667                                                           ForesterUtil.LINE_SEPARATOR,
 668                                                           null );
 669             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 670                                                                CharacterStateMatrix.GainLossStates.LOSS,
 671                                                                outfile_name
 672                                                                        + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_BC,
 673                                                                sep,
 674                                                                ForesterUtil.LINE_SEPARATOR,
 675                                                                null );
 676             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 677                     + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC, sep, ForesterUtil.LINE_SEPARATOR, null );
 678             if ( all_binary_domains_combination_gained_fitch != null ) {
 679                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 680                                                                                     dc_type,
 681                                                                                     all_binary_domains_combination_gained_fitch,
 682                                                                                     true );
 683             }
 684             if ( all_binary_domains_combination_lost_fitch != null ) {
 685                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 686                                                                                     dc_type,
 687                                                                                     all_binary_domains_combination_lost_fitch,
 688                                                                                     false );
 689             }
 690             if ( output_binary_domain_combinations_for_graphs ) {
 691                 SurfacingUtil
 692                         .writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( domain_parsimony
 693                                                                                                            .getGainLossMatrix(),
 694                                                                                                    null,
 695                                                                                                    outfile_name
 696                                                                                                            + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS,
 697                                                                                                    sep,
 698                                                                                                    ForesterUtil.LINE_SEPARATOR,
 699                                                                                                    BinaryDomainCombination.OutputFormat.DOT );
 700             }
 701             // HTML:
 702             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 703                                            go_id_to_term_map,
 704                                            go_namespace_limit,
 705                                            true,
 706                                            domain_parsimony.getGainLossMatrix(),
 707                                            CharacterStateMatrix.GainLossStates.GAIN,
 708                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC,
 709                                            sep,
 710                                            ForesterUtil.LINE_SEPARATOR,
 711                                            "Fitch Parsimony | Gains | Domain Combinations",
 712                                            "+",
 713                                            null,
 714                                            all_pfams_encountered,
 715                                            all_pfams_gained_as_dom_combinations,
 716                                            "_fitch_gains_dc",
 717                                            tax_code_to_id_map );
 718             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 719                                            go_id_to_term_map,
 720                                            go_namespace_limit,
 721                                            true,
 722                                            domain_parsimony.getGainLossMatrix(),
 723                                            CharacterStateMatrix.GainLossStates.LOSS,
 724                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC,
 725                                            sep,
 726                                            ForesterUtil.LINE_SEPARATOR,
 727                                            "Fitch Parsimony | Losses | Domain Combinations",
 728                                            "-",
 729                                            null,
 730                                            all_pfams_encountered,
 731                                            all_pfams_lost_as_dom_combinations,
 732                                            "_fitch_losses_dc",
 733                                            tax_code_to_id_map );
 734             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 735                                            go_id_to_term_map,
 736                                            go_namespace_limit,
 737                                            true,
 738                                            domain_parsimony.getGainLossMatrix(),
 739                                            null,
 740                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC,
 741                                            sep,
 742                                            ForesterUtil.LINE_SEPARATOR,
 743                                            "Fitch Parsimony | Present | Domain Combinations",
 744                                            "",
 745                                            null,
 746                                            all_pfams_encountered,
 747                                            null,
 748                                            "_fitch_present_dc",
 749                                            tax_code_to_id_map );
 750             writeAllEncounteredPfamsToFile( domain_id_to_go_ids_map,
 751                                             go_id_to_term_map,
 752                                             outfile_name,
 753                                             all_pfams_encountered );
 754             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX, all_pfams_gained_as_domains );
 755             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX, all_pfams_lost_as_domains );
 756             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DC_SUFFIX,
 757                               all_pfams_gained_as_dom_combinations );
 758             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DC_SUFFIX, all_pfams_lost_as_dom_combinations );
 759             preparePhylogeny( local_phylogeny_l,
 760                               domain_parsimony,
 761                               date_time,
 762                               "Fitch parsimony on binary domain combination presence/absence randomization: "
 763                                       + randomization,
 764                               "fitch_on_binary_domain_combinations_" + outfile_name,
 765                               parameters_str );
 766             SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 767                     + surfacing.BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH );
 768             calculateIndependentDomainCombinationGains( local_phylogeny_l,
 769                                                         outfile_name
 770                                                                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX,
 771                                                         outfile_name
 772                                                                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX,
 773                                                         outfile_name
 774                                                                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX,
 775                                                         outfile_name
 776                                                                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX,
 777                                                         outfile_name + "_indep_dc_gains_fitch_lca_ranks.txt",
 778                                                         outfile_name + "_indep_dc_gains_fitch_lca_taxonomies.txt",
 779                                                         outfile_name + "_indep_dc_gains_fitch_protein_statistics.txt",
 780                                                         protein_length_stats_by_dc,
 781                                                         domain_number_stats_by_dc,
 782                                                         domain_length_stats_by_domain );
 783         }
 784     }
 785
 786     public static void executeParsimonyAnalysisForSecondaryFeatures( final String outfile_name,
 787                                                                      final DomainParsimonyCalculator secondary_features_parsimony,
 788                                                                      final Phylogeny phylogeny,
 789                                                                      final String parameters_str,
 790                                                                      final Map<Species, MappingResults> mapping_results_map ) {
 791         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 792         final String date_time = ForesterUtil.getCurrentDateTime();
 793         System.out.println();
 794         writeToNexus( outfile_name + surfacing.NEXUS_SECONDARY_FEATURES,
 795                       secondary_features_parsimony.createMatrixOfSecondaryFeaturePresenceOrAbsence( null ),
 796                       phylogeny );
 797         Phylogeny local_phylogeny_copy = phylogeny.copy();
 798         secondary_features_parsimony.executeDolloParsimonyOnSecondaryFeatures( mapping_results_map );
 799         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossMatrix(), outfile_name
 800                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 801         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossCountsMatrix(), outfile_name
 802                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 803         SurfacingUtil
 804                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 805                                                       CharacterStateMatrix.GainLossStates.GAIN,
 806                                                       outfile_name
 807                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES,
 808                                                       sep,
 809                                                       ForesterUtil.LINE_SEPARATOR,
 810                                                       null );
 811         SurfacingUtil
 812                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 813                                                       CharacterStateMatrix.GainLossStates.LOSS,
 814                                                       outfile_name
 815                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES,
 816                                                       sep,
 817                                                       ForesterUtil.LINE_SEPARATOR,
 818                                                       null );
 819         SurfacingUtil
 820                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 821                                                       null,
 822                                                       outfile_name
 823                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES,
 824                                                       sep,
 825                                                       ForesterUtil.LINE_SEPARATOR,
 826                                                       null );
 827         preparePhylogeny( local_phylogeny_copy,
 828                           secondary_features_parsimony,
 829                           date_time,
 830                           "Dollo parsimony on secondary feature presence/absence",
 831                           "dollo_on_secondary_features_" + outfile_name,
 832                           parameters_str );
 833         SurfacingUtil.writePhylogenyToFile( local_phylogeny_copy, outfile_name
 834                 + surfacing.SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 835         // FITCH DOMAIN COMBINATIONS
 836         // -------------------------
 837         local_phylogeny_copy = phylogeny.copy();
 838         final String randomization = "no";
 839         secondary_features_parsimony.executeFitchParsimonyOnBinaryDomainCombintionOnSecondaryFeatures( USE_LAST );
 840         preparePhylogeny( local_phylogeny_copy,
 841                           secondary_features_parsimony,
 842                           date_time,
 843                           "Fitch parsimony on secondary binary domain combination presence/absence randomization: "
 844                                   + randomization,
 845                           "fitch_on_binary_domain_combinations_" + outfile_name,
 846                           parameters_str );
 847         SurfacingUtil.writePhylogenyToFile( local_phylogeny_copy, outfile_name
 848                 + surfacing.BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH_MAPPED );
 849         calculateIndependentDomainCombinationGains( local_phylogeny_copy, outfile_name
 850                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_MAPPED_OUTPUT_SUFFIX, outfile_name
 851                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_MAPPED_OUTPUT_SUFFIX, outfile_name
 852                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_SUFFIX, outfile_name
 853                 + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_UNIQUE_SUFFIX, outfile_name
 854                 + "_MAPPED_indep_dc_gains_fitch_lca_ranks.txt", outfile_name
 855                 + "_MAPPED_indep_dc_gains_fitch_lca_taxonomies.txt", null, null, null, null );
 856     }
 857
 858     public static void extractProteinNames( final List<Protein> proteins,
 859                                             final List<DomainId> query_domain_ids_nc_order,
 860                                             final Writer out,
 861                                             final String separator,
 862                                             final String limit_to_species ) throws IOException {
 863         for( final Protein protein : proteins ) {
 864             if ( ForesterUtil.isEmpty( limit_to_species )
 865                     || protein.getSpecies().getSpeciesId().equalsIgnoreCase( limit_to_species ) ) {
 866                 if ( protein.contains( query_domain_ids_nc_order, true ) ) {
 867                     out.write( protein.getSpecies().getSpeciesId() );
 868                     out.write( separator );
 869                     out.write( protein.getProteinId().getId() );
 870                     out.write( separator );
 871                     out.write( "[" );
 872                     final Set<DomainId> visited_domain_ids = new HashSet<DomainId>();
 873                     boolean first = true;
 874                     for( final Domain domain : protein.getProteinDomains() ) {
 875                         if ( !visited_domain_ids.contains( domain.getDomainId() ) ) {
 876                             visited_domain_ids.add( domain.getDomainId() );
 877                             if ( first ) {
 878                                 first = false;
 879                             }
 880                             else {
 881                                 out.write( " " );
 882                             }
 883                             out.write( domain.getDomainId().getId() );
 884                             out.write( " {" );
 885                             out.write( "" + domain.getTotalCount() );
 886                             out.write( "}" );
 887                         }
 888                     }
 889                     out.write( "]" );
 890                     out.write( separator );
 891                     if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 892                             .equals( SurfacingConstants.NONE ) ) ) {
 893                         out.write( protein.getDescription() );
 894                     }
 895                     out.write( separator );
 896                     if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 897                             .equals( SurfacingConstants.NONE ) ) ) {
 898                         out.write( protein.getAccession() );
 899                     }
 900                     out.write( SurfacingConstants.NL );
 901                 }
 902             }
 903         }
 904         out.flush();
 905     }
 906
 907     public static void extractProteinNames( final SortedMap<Species, List<Protein>> protein_lists_per_species,
 908                                             final DomainId domain_id,
 909                                             final Writer out,
 910                                             final String separator,
 911                                             final String limit_to_species,
 912                                             final double domain_e_cutoff ) throws IOException {
 913         System.out.println( "Per domain E-value: " + domain_e_cutoff );
 914         for( final Species species : protein_lists_per_species.keySet() ) {
 915             System.out.println( species + ":" );
 916             for( final Protein protein : protein_lists_per_species.get( species ) ) {
 917                 if ( ForesterUtil.isEmpty( limit_to_species )
 918                         || protein.getSpecies().getSpeciesId().equalsIgnoreCase( limit_to_species ) ) {
 919                     final List<Domain> domains = protein.getProteinDomains( domain_id );
 920                     if ( domains.size() > 0 ) {
 921                         out.write( protein.getSpecies().getSpeciesId() );
 922                         out.write( separator );
 923                         out.write( protein.getProteinId().getId() );
 924                         out.write( separator );
 925                         out.write( domain_id.toString() );
 926                         out.write( separator );
 927                         int prev_to = -1;
 928                         for( final Domain domain : domains ) {
 929                             if ( ( domain_e_cutoff < 0 ) || ( domain.getPerDomainEvalue() <= domain_e_cutoff ) ) {
 930                                 out.write( "/" );
 931                                 out.write( domain.getFrom() + "-" + domain.getTo() );
 932                                 if ( prev_to >= 0 ) {
 933                                     final int l = domain.getFrom() - prev_to;
 934                                     System.out.println( l );
 935                                 }
 936                                 prev_to = domain.getTo();
 937                             }
 938                         }
 939                         out.write( "/" );
 940                         out.write( separator );
 941                         final List<Domain> domain_list = new ArrayList<Domain>();
 942                         for( final Domain domain : protein.getProteinDomains() ) {
 943                             if ( ( domain_e_cutoff < 0 ) || ( domain.getPerDomainEvalue() <= domain_e_cutoff ) ) {
 944                                 domain_list.add( domain );
 945                             }
 946                         }
 947                         final Domain domain_ary[] = new Domain[ domain_list.size() ];
 948                         for( int i = 0; i < domain_list.size(); ++i ) {
 949                             domain_ary[ i ] = domain_list.get( i );
 950                         }
 951                         Arrays.sort( domain_ary, new DomainComparator( true ) );
 952                         out.write( "{" );
 953                         boolean first = true;
 954                         for( final Domain domain : domain_ary ) {
 955                             if ( first ) {
 956                                 first = false;
 957                             }
 958                             else {
 959                                 out.write( "," );
 960                             }
 961                             out.write( domain.getDomainId().toString() );
 962                             out.write( ":" + domain.getFrom() + "-" + domain.getTo() );
 963                             out.write( ":" + domain.getPerDomainEvalue() );
 964                         }
 965                         out.write( "}" );
 966                         if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 967                                 .equals( SurfacingConstants.NONE ) ) ) {
 968                             out.write( protein.getDescription() );
 969                         }
 970                         out.write( separator );
 971                         if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 972                                 .equals( SurfacingConstants.NONE ) ) ) {
 973                             out.write( protein.getAccession() );
 974                         }
 975                         out.write( SurfacingConstants.NL );
 976                     }
 977                 }
 978             }
 979         }
 980         out.flush();
 981     }
 982
 983     public static SortedSet<DomainId> getAllDomainIds( final List<GenomeWideCombinableDomains> gwcd_list ) {
 984         final SortedSet<DomainId> all_domains_ids = new TreeSet<DomainId>();
 985         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
 986             final Set<DomainId> all_domains = gwcd.getAllDomainIds();
 987             //    for( final Domain domain : all_domains ) {
 988             all_domains_ids.addAll( all_domains );
 989             //    }
 990         }
 991         return all_domains_ids;
 992     }
 993
 994     public static SortedMap<String, Integer> getDomainCounts( final List<Protein> protein_domain_collections ) {
 995         final SortedMap<String, Integer> map = new TreeMap<String, Integer>();
 996         for( final Protein protein_domain_collection : protein_domain_collections ) {
 997             for( final Object name : protein_domain_collection.getProteinDomains() ) {
 998                 final BasicDomain protein_domain = ( BasicDomain ) name;
 999                 final String id = protein_domain.getDomainId().getId();
1000                 if ( map.containsKey( id ) ) {
1001                     map.put( id, map.get( id ) + 1 );
1002                 }
1003                 else {
1004                     map.put( id, 1 );
1005                 }
1006             }
1007         }
1008         return map;
1009     }
1010
1011     public static int getNumberOfNodesLackingName( final Phylogeny p, final StringBuilder names ) {
1012         final PhylogenyNodeIterator it = p.iteratorPostorder();
1013         int c = 0;
1014         while ( it.hasNext() ) {
1015             final PhylogenyNode n = it.next();
1016             if ( ForesterUtil.isEmpty( n.getName() )
1017                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
1018                             .getScientificName() ) )
1019                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
1020                             .getCommonName() ) ) ) {
1021                 if ( n.getParent() != null ) {
1022                     names.append( " " );
1023                     names.append( n.getParent().getName() );
1024                 }
1025                 final List l = n.getAllExternalDescendants();
1026                 for( final Object object : l ) {
1027                     System.out.println( l.toString() );
1028                 }
1029                 ++c;
1030             }
1031         }
1032         return c;
1033     }
1034
1035     /**
1036      * Returns true is Domain domain falls in an uninterrupted stretch of
1037      * covered positions.
1038      *
1039      * @param domain
1040      * @param covered_positions
1041      * @return
1042      */
1043     public static boolean isEngulfed( final Domain domain, final List<Boolean> covered_positions ) {
1044         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
1045             if ( ( i >= covered_positions.size() ) || ( covered_positions.get( i ) != true ) ) {
1046                 return false;
1047             }
1048         }
1049         return true;
1050     }
1051
1052     public static void performDomainArchitectureAnalysis( final SortedMap<String, Set<String>> domain_architecutures,
1053                                                           final SortedMap<String, Integer> domain_architecuture_counts,
1054                                                           final int min_count,
1055                                                           final File da_counts_outfile,
1056                                                           final File unique_da_outfile ) {
1057         checkForOutputFileWriteability( da_counts_outfile );
1058         checkForOutputFileWriteability( unique_da_outfile );
1059         try {
1060             final BufferedWriter da_counts_out = new BufferedWriter( new FileWriter( da_counts_outfile ) );
1061             final BufferedWriter unique_da_out = new BufferedWriter( new FileWriter( unique_da_outfile ) );
1062             final Iterator<Entry<String, Integer>> it = domain_architecuture_counts.entrySet().iterator();
1063             while ( it.hasNext() ) {
1064                 final Map.Entry<String, Integer> e = it.next();
1065                 final String da = e.getKey();
1066                 final int count = e.getValue();
1067                 if ( count >= min_count ) {
1068                     da_counts_out.write( da );
1069                     da_counts_out.write( "\t" );
1070                     da_counts_out.write( String.valueOf( count ) );
1071                     da_counts_out.write( ForesterUtil.LINE_SEPARATOR );
1072                 }
1073                 if ( count == 1 ) {
1074                     final Iterator<Entry<String, Set<String>>> it2 = domain_architecutures.entrySet().iterator();
1075                     while ( it2.hasNext() ) {
1076                         final Map.Entry<String, Set<String>> e2 = it2.next();
1077                         final String genome = e2.getKey();
1078                         final Set<String> das = e2.getValue();
1079                         if ( das.contains( da ) ) {
1080                             unique_da_out.write( genome );
1081                             unique_da_out.write( "\t" );
1082                             unique_da_out.write( da );
1083                             unique_da_out.write( ForesterUtil.LINE_SEPARATOR );
1084                         }
1085                     }
1086                 }
1087             }
1088             unique_da_out.close();
1089             da_counts_out.close();
1090         }
1091         catch ( final IOException e ) {
1092             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1093         }
1094         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote distance matrices to \"" + da_counts_outfile + "\"" );
1095         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote distance matrices to \"" + unique_da_outfile + "\"" );
1096         //
1097     }
1098
1099     public static void preparePhylogeny( final Phylogeny p,
1100                                          final DomainParsimonyCalculator domain_parsimony,
1101                                          final String date_time,
1102                                          final String method,
1103                                          final String name,
1104                                          final String parameters_str ) {
1105         domain_parsimony.decoratePhylogenyWithDomains( p );
1106         final StringBuilder desc = new StringBuilder();
1107         desc.append( "[Method: " + method + "] [Date: " + date_time + "] " );
1108         desc.append( "[Cost: " + domain_parsimony.getCost() + "] " );
1109         desc.append( "[Gains: " + domain_parsimony.getTotalGains() + "] " );
1110         desc.append( "[Losses: " + domain_parsimony.getTotalLosses() + "] " );
1111         desc.append( "[Unchanged: " + domain_parsimony.getTotalUnchanged() + "] " );
1112         desc.append( "[Parameters: " + parameters_str + "]" );
1113         p.setName( name );
1114         p.setDescription( desc.toString() );
1115         p.setConfidence( new Confidence( domain_parsimony.getCost(), "parsimony" ) );
1116         p.setRerootable( false );
1117         p.setRooted( true );
1118     }
1119
1120     /*
1121      * species | protein id | n-terminal domain | c-terminal domain | n-terminal domain per domain E-value | c-terminal domain per domain E-value
1122      *
1123      *
1124      */
1125     static public StringBuffer proteinToDomainCombinations( final Protein protein,
1126                                                             final String protein_id,
1127                                                             final String separator ) {
1128         final StringBuffer sb = new StringBuffer();
1129         if ( protein.getSpecies() == null ) {
1130             throw new IllegalArgumentException( "species must not be null" );
1131         }
1132         if ( ForesterUtil.isEmpty( protein.getSpecies().getSpeciesId() ) ) {
1133             throw new IllegalArgumentException( "species id must not be empty" );
1134         }
1135         final List<Domain> domains = protein.getProteinDomains();
1136         if ( domains.size() > 1 ) {
1137             final Map<String, Integer> counts = new HashMap<String, Integer>();
1138             for( final Domain domain : domains ) {
1139                 final String id = domain.getDomainId().getId();
1140                 if ( counts.containsKey( id ) ) {
1141                     counts.put( id, counts.get( id ) + 1 );
1142                 }
1143                 else {
1144                     counts.put( id, 1 );
1145                 }
1146             }
1147             final Set<String> dcs = new HashSet<String>();
1148             for( int i = 1; i < domains.size(); ++i ) {
1149                 for( int j = 0; j < i; ++j ) {
1150                     Domain domain_n = domains.get( i );
1151                     Domain domain_c = domains.get( j );
1152                     if ( domain_n.getFrom() > domain_c.getFrom() ) {
1153                         domain_n = domains.get( j );
1154                         domain_c = domains.get( i );
1155                     }
1156                     final String dc = domain_n.getDomainId().getId() + domain_c.getDomainId().getId();
1157                     if ( !dcs.contains( dc ) ) {
1158                         dcs.add( dc );
1159                         sb.append( protein.getSpecies() );
1160                         sb.append( separator );
1161                         sb.append( protein_id );
1162                         sb.append( separator );
1163                         sb.append( domain_n.getDomainId().getId() );
1164                         sb.append( separator );
1165                         sb.append( domain_c.getDomainId().getId() );
1166                         sb.append( separator );
1167                         sb.append( domain_n.getPerDomainEvalue() );
1168                         sb.append( separator );
1169                         sb.append( domain_c.getPerDomainEvalue() );
1170                         sb.append( separator );
1171                         sb.append( counts.get( domain_n.getDomainId().getId() ) );
1172                         sb.append( separator );
1173                         sb.append( counts.get( domain_c.getDomainId().getId() ) );
1174                         sb.append( ForesterUtil.LINE_SEPARATOR );
1175                     }
1176                 }
1177             }
1178         }
1179         else if ( domains.size() == 1 ) {
1180             sb.append( protein.getSpecies() );
1181             sb.append( separator );
1182             sb.append( protein_id );
1183             sb.append( separator );
1184             sb.append( domains.get( 0 ).getDomainId().getId() );
1185             sb.append( separator );
1186             sb.append( separator );
1187             sb.append( domains.get( 0 ).getPerDomainEvalue() );
1188             sb.append( separator );
1189             sb.append( separator );
1190             sb.append( 1 );
1191             sb.append( separator );
1192             sb.append( ForesterUtil.LINE_SEPARATOR );
1193         }
1194         else {
1195             sb.append( protein.getSpecies() );
1196             sb.append( separator );
1197             sb.append( protein_id );
1198             sb.append( separator );
1199             sb.append( separator );
1200             sb.append( separator );
1201             sb.append( separator );
1202             sb.append( separator );
1203             sb.append( separator );
1204             sb.append( ForesterUtil.LINE_SEPARATOR );
1205         }
1206         return sb;
1207     }
1208
1209     /**
1210      *
1211      * Example regarding engulfment: ------------0.1 ----------0.2 --0.3 =>
1212      * domain with 0.3 is ignored
1213      *
1214      * -----------0.1 ----------0.2 --0.3 => domain with 0.3 is ignored
1215      *
1216      *
1217      * ------------0.1 ----------0.3 --0.2 => domains with 0.3 and 0.2 are _not_
1218      * ignored
1219      *
1220      * @param max_allowed_overlap
1221      *            maximal allowed overlap (inclusive) to be still considered not
1222      *            overlapping (zero or negative value to allow any overlap)
1223      * @param remove_engulfed_domains
1224      *            to remove domains which are completely engulfed by coverage of
1225      *            domains with better support
1226      * @param protein
1227      * @return
1228      */
1229     public static Protein removeOverlappingDomains( final int max_allowed_overlap,
1230                                                     final boolean remove_engulfed_domains,
1231                                                     final Protein protein ) {
1232         final Protein pruned_protein = new BasicProtein( protein.getProteinId().getId(), protein.getSpecies()
1233                 .getSpeciesId(), protein.getLength() );
1234         final List<Domain> sorted = SurfacingUtil.sortDomainsWithAscendingConfidenceValues( protein );
1235         final List<Boolean> covered_positions = new ArrayList<Boolean>();
1236         for( final Domain domain : sorted ) {
1237             if ( ( ( max_allowed_overlap < 0 ) || ( SurfacingUtil.calculateOverlap( domain, covered_positions ) <= max_allowed_overlap ) )
1238                     && ( !remove_engulfed_domains || !isEngulfed( domain, covered_positions ) ) ) {
1239                 final int covered_positions_size = covered_positions.size();
1240                 for( int i = covered_positions_size; i < domain.getFrom(); ++i ) {
1241                     covered_positions.add( false );
1242                 }
1243                 final int new_covered_positions_size = covered_positions.size();
1244                 for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
1245                     if ( i < new_covered_positions_size ) {
1246                         covered_positions.set( i, true );
1247                     }
1248                     else {
1249                         covered_positions.add( true );
1250                     }
1251                 }
1252                 pruned_protein.addProteinDomain( domain );
1253             }
1254         }
1255         return pruned_protein;
1256     }
1257
1258     public static List<Domain> sortDomainsWithAscendingConfidenceValues( final Protein protein ) {
1259         final List<Domain> domains = new ArrayList<Domain>();
1260         for( final Domain d : protein.getProteinDomains() ) {
1261             domains.add( d );
1262         }
1263         Collections.sort( domains, SurfacingUtil.ASCENDING_CONFIDENCE_VALUE_ORDER );
1264         return domains;
1265     }
1266
1267     public static int storeDomainArchitectures( final String genome,
1268                                                 final SortedMap<String, Set<String>> domain_architecutures,
1269                                                 final List<Protein> protein_list,
1270                                                 final Map<String, Integer> distinct_domain_architecuture_counts ) {
1271         final Set<String> da = new HashSet<String>();
1272         domain_architecutures.put( genome, da );
1273         for( final Protein protein : protein_list ) {
1274             final String da_str = ( ( BasicProtein ) protein ).toDomainArchitectureString( "~", 3, "=" );
1275             if ( !da.contains( da_str ) ) {
1276                 if ( !distinct_domain_architecuture_counts.containsKey( da_str ) ) {
1277                     distinct_domain_architecuture_counts.put( da_str, 1 );
1278                 }
1279                 else {
1280                     distinct_domain_architecuture_counts.put( da_str,
1281                                                               distinct_domain_architecuture_counts.get( da_str ) + 1 );
1282                 }
1283                 da.add( da_str );
1284             }
1285         }
1286         return da.size();
1287     }
1288
1289     public static void writeAllDomainsChangedOnAllSubtrees( final Phylogeny p,
1290                                                             final boolean get_gains,
1291                                                             final String outdir,
1292                                                             final String suffix_for_filename ) throws IOException {
1293         CharacterStateMatrix.GainLossStates state = CharacterStateMatrix.GainLossStates.GAIN;
1294         if ( !get_gains ) {
1295             state = CharacterStateMatrix.GainLossStates.LOSS;
1296         }
1297         final File base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES,
1298                                                                   false,
1299                                                                   state,
1300                                                                   outdir );
1301         for( final PhylogenyNodeIterator it = p.iteratorPostorder(); it.hasNext(); ) {
1302             final PhylogenyNode node = it.next();
1303             if ( !node.isExternal() ) {
1304                 final SortedSet<String> domains = collectAllDomainsChangedOnSubtree( node, get_gains );
1305                 if ( domains.size() > 0 ) {
1306                     final Writer writer = ForesterUtil.createBufferedWriter( base_dir + ForesterUtil.FILE_SEPARATOR
1307                             + node.getName() + suffix_for_filename );
1308                     for( final String domain : domains ) {
1309                         writer.write( domain );
1310                         writer.write( ForesterUtil.LINE_SEPARATOR );
1311                     }
1312                     writer.close();
1313                 }
1314             }
1315         }
1316     }
1317
1318     public static void writeBinaryDomainCombinationsFileForGraphAnalysis( final String[][] input_file_properties,
1319                                                                           final File output_dir,
1320                                                                           final GenomeWideCombinableDomains gwcd,
1321                                                                           final int i,
1322                                                                           final GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1323         File dc_outfile_dot = new File( input_file_properties[ i ][ 1 ]
1324                 + surfacing.DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS );
1325         if ( output_dir != null ) {
1326             dc_outfile_dot = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile_dot );
1327         }
1328         checkForOutputFileWriteability( dc_outfile_dot );
1329         final SortedSet<BinaryDomainCombination> binary_combinations = createSetOfAllBinaryDomainCombinationsPerGenome( gwcd );
1330         try {
1331             final BufferedWriter out_dot = new BufferedWriter( new FileWriter( dc_outfile_dot ) );
1332             for( final BinaryDomainCombination bdc : binary_combinations ) {
1333                 out_dot.write( bdc.toGraphDescribingLanguage( BinaryDomainCombination.OutputFormat.DOT, null, null )
1334                         .toString() );
1335                 out_dot.write( SurfacingConstants.NL );
1336             }
1337             out_dot.close();
1338         }
1339         catch ( final IOException e ) {
1340             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1341         }
1342         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote binary domain combination for \""
1343                 + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1344                 + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile_dot + "\"" );
1345     }
1346
1347     public static void writeBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1348                                                             final CharacterStateMatrix.GainLossStates state,
1349                                                             final String filename,
1350                                                             final String indentifier_characters_separator,
1351                                                             final String character_separator,
1352                                                             final Map<String, String> descriptions ) {
1353         final File outfile = new File( filename );
1354         checkForOutputFileWriteability( outfile );
1355         final SortedSet<String> sorted_ids = new TreeSet<String>();
1356         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1357             sorted_ids.add( matrix.getIdentifier( i ) );
1358         }
1359         try {
1360             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1361             for( final String id : sorted_ids ) {
1362                 out.write( indentifier_characters_separator );
1363                 out.write( "#" + id );
1364                 out.write( indentifier_characters_separator );
1365                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1366                     // Not nice:
1367                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1368                     if ( ( matrix.getState( id, c ) == state )
1369                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1370                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1371                         out.write( matrix.getCharacter( c ) );
1372                         if ( ( descriptions != null ) && !descriptions.isEmpty()
1373                                 && descriptions.containsKey( matrix.getCharacter( c ) ) ) {
1374                             out.write( "\t" );
1375                             out.write( descriptions.get( matrix.getCharacter( c ) ) );
1376                         }
1377                         out.write( character_separator );
1378                     }
1379                 }
1380             }
1381             out.flush();
1382             out.close();
1383         }
1384         catch ( final IOException e ) {
1385             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1386         }
1387         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1388     }
1389
1390     public static void writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1391                                                                                                  final CharacterStateMatrix.GainLossStates state,
1392                                                                                                  final String filename,
1393                                                                                                  final String indentifier_characters_separator,
1394                                                                                                  final String character_separator,
1395                                                                                                  final BinaryDomainCombination.OutputFormat bc_output_format ) {
1396         final File outfile = new File( filename );
1397         checkForOutputFileWriteability( outfile );
1398         final SortedSet<String> sorted_ids = new TreeSet<String>();
1399         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1400             sorted_ids.add( matrix.getIdentifier( i ) );
1401         }
1402         try {
1403             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1404             for( final String id : sorted_ids ) {
1405                 out.write( indentifier_characters_separator );
1406                 out.write( "#" + id );
1407                 out.write( indentifier_characters_separator );
1408                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1409                     // Not nice:
1410                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1411                     if ( ( matrix.getState( id, c ) == state )
1412                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1413                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1414                         BinaryDomainCombination bdc = null;
1415                         try {
1416                             bdc = BasicBinaryDomainCombination.createInstance( matrix.getCharacter( c ) );
1417                         }
1418                         catch ( final Exception e ) {
1419                             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
1420                         }
1421                         out.write( bdc.toGraphDescribingLanguage( bc_output_format, null, null ).toString() );
1422                         out.write( character_separator );
1423                     }
1424                 }
1425             }
1426             out.flush();
1427             out.close();
1428         }
1429         catch ( final IOException e ) {
1430             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1431         }
1432         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1433     }
1434
1435     public static void writeBinaryStatesMatrixToList( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1436                                                       final Map<GoId, GoTerm> go_id_to_term_map,
1437                                                       final GoNameSpace go_namespace_limit,
1438                                                       final boolean domain_combinations,
1439                                                       final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1440                                                       final CharacterStateMatrix.GainLossStates state,
1441                                                       final String filename,
1442                                                       final String indentifier_characters_separator,
1443                                                       final String character_separator,
1444                                                       final String title_for_html,
1445                                                       final String prefix_for_html,
1446                                                       final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1447                                                       final SortedSet<String> all_pfams_encountered,
1448                                                       final SortedSet<String> pfams_gained_or_lost,
1449                                                       final String suffix_for_per_node_events_file,
1450                                                       final Map<String, Integer> tax_code_to_id_map ) {
1451         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1452             throw new IllegalArgumentException( "attempt to use GO namespace limit without a GO-id to term map" );
1453         }
1454         else if ( ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) ) {
1455             throw new IllegalArgumentException( "attempt to output detailed HTML without a Pfam to GO map" );
1456         }
1457         else if ( ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1458             throw new IllegalArgumentException( "attempt to output detailed HTML without a GO-id to term map" );
1459         }
1460         final File outfile = new File( filename );
1461         checkForOutputFileWriteability( outfile );
1462         final SortedSet<String> sorted_ids = new TreeSet<String>();
1463         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1464             sorted_ids.add( matrix.getIdentifier( i ) );
1465         }
1466         try {
1467             final Writer out = new BufferedWriter( new FileWriter( outfile ) );
1468             final File per_node_go_mapped_domain_gain_loss_files_base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES,
1469                                                                                                                 domain_combinations,
1470                                                                                                                 state,
1471                                                                                                                 filename );
1472             Writer per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1473             File per_node_go_mapped_domain_gain_loss_outfile = null;
1474             int per_node_counter = 0;
1475             out.write( "<html>" );
1476             out.write( SurfacingConstants.NL );
1477             addHtmlHead( out, title_for_html );
1478             out.write( SurfacingConstants.NL );
1479             out.write( "<body>" );
1480             out.write( SurfacingConstants.NL );
1481             out.write( "<h1>" );
1482             out.write( SurfacingConstants.NL );
1483             out.write( title_for_html );
1484             out.write( SurfacingConstants.NL );
1485             out.write( "</h1>" );
1486             out.write( SurfacingConstants.NL );
1487             out.write( "<table>" );
1488             out.write( SurfacingConstants.NL );
1489             for( final String id : sorted_ids ) {
1490                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1491                 if ( matcher.matches() ) {
1492                     continue;
1493                 }
1494                 out.write( "<tr>" );
1495                 out.write( "<td>" );
1496                 out.write( "<a href=\"#" + id + "\">" + id + "</a>" );
1497                 out.write( "</td>" );
1498                 out.write( "</tr>" );
1499                 out.write( SurfacingConstants.NL );
1500             }
1501             out.write( "</table>" );
1502             out.write( SurfacingConstants.NL );
1503             for( final String id : sorted_ids ) {
1504                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1505                 if ( matcher.matches() ) {
1506                     continue;
1507                 }
1508                 out.write( SurfacingConstants.NL );
1509                 out.write( "<h2>" );
1510                 out.write( "<a name=\"" + id + "\">" + id + "</a>" );
1511                 writeTaxonomyLinks( out, id, tax_code_to_id_map );
1512                 out.write( "</h2>" );
1513                 out.write( SurfacingConstants.NL );
1514                 out.write( "<table>" );
1515                 out.write( SurfacingConstants.NL );
1516                 out.write( "<tr>" );
1517                 out.write( "<td><b>" );
1518                 out.write( "Pfam domain(s)" );
1519                 out.write( "</b></td><td><b>" );
1520                 out.write( "GO term acc" );
1521                 out.write( "</b></td><td><b>" );
1522                 out.write( "GO term" );
1523                 out.write( "</b></td><td><b>" );
1524                 out.write( "GO namespace" );
1525                 out.write( "</b></td>" );
1526                 out.write( "</tr>" );
1527                 out.write( SurfacingConstants.NL );
1528                 out.write( "</tr>" );
1529                 out.write( SurfacingConstants.NL );
1530                 per_node_counter = 0;
1531                 if ( matrix.getNumberOfCharacters() > 0 ) {
1532                     per_node_go_mapped_domain_gain_loss_outfile = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
1533                             + ForesterUtil.FILE_SEPARATOR + id + suffix_for_per_node_events_file );
1534                     SurfacingUtil.checkForOutputFileWriteability( per_node_go_mapped_domain_gain_loss_outfile );
1535                     per_node_go_mapped_domain_gain_loss_outfile_writer = ForesterUtil
1536                             .createBufferedWriter( per_node_go_mapped_domain_gain_loss_outfile );
1537                 }
1538                 else {
1539                     per_node_go_mapped_domain_gain_loss_outfile = null;
1540                     per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1541                 }
1542                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1543                     // Not nice:
1544                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1545                     if ( ( matrix.getState( id, c ) == state )
1546                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) || ( matrix
1547                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) ) ) ) {
1548                         final String character = matrix.getCharacter( c );
1549                         String domain_0 = "";
1550                         String domain_1 = "";
1551                         if ( character.indexOf( BinaryDomainCombination.SEPARATOR ) > 0 ) {
1552                             final String[] s = character.split( BinaryDomainCombination.SEPARATOR );
1553                             if ( s.length != 2 ) {
1554                                 throw new AssertionError( "this should not have happened: unexpected format for domain combination: ["
1555                                         + character + "]" );
1556                             }
1557                             domain_0 = s[ 0 ];
1558                             domain_1 = s[ 1 ];
1559                         }
1560                         else {
1561                             domain_0 = character;
1562                         }
1563                         writeDomainData( domain_id_to_go_ids_map,
1564                                          go_id_to_term_map,
1565                                          go_namespace_limit,
1566                                          out,
1567                                          domain_0,
1568                                          domain_1,
1569                                          prefix_for_html,
1570                                          character_separator,
1571                                          domain_id_to_secondary_features_maps,
1572                                          null );
1573                         all_pfams_encountered.add( domain_0 );
1574                         if ( pfams_gained_or_lost != null ) {
1575                             pfams_gained_or_lost.add( domain_0 );
1576                         }
1577                         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
1578                             all_pfams_encountered.add( domain_1 );
1579                             if ( pfams_gained_or_lost != null ) {
1580                                 pfams_gained_or_lost.add( domain_1 );
1581                             }
1582                         }
1583                         if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1584                             writeDomainsToIndividualFilePerTreeNode( per_node_go_mapped_domain_gain_loss_outfile_writer,
1585                                                                      domain_0,
1586                                                                      domain_1 );
1587                             per_node_counter++;
1588                         }
1589                     }
1590                 }
1591                 if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1592                     per_node_go_mapped_domain_gain_loss_outfile_writer.close();
1593                     if ( per_node_counter < 1 ) {
1594                         per_node_go_mapped_domain_gain_loss_outfile.delete();
1595                     }
1596                     per_node_counter = 0;
1597                 }
1598                 out.write( "</table>" );
1599                 out.write( SurfacingConstants.NL );
1600                 out.write( "<hr>" );
1601                 out.write( SurfacingConstants.NL );
1602             } // for( final String id : sorted_ids ) {
1603             out.write( "</body>" );
1604             out.write( SurfacingConstants.NL );
1605             out.write( "</html>" );
1606             out.write( SurfacingConstants.NL );
1607             out.flush();
1608             out.close();
1609         }
1610         catch ( final IOException e ) {
1611             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1612         }
1613         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters detailed HTML list: \"" + filename + "\"" );
1614     }
1615
1616     public static void writeDomainCombinationsCountsFile( final String[][] input_file_properties,
1617                                                           final File output_dir,
1618                                                           final Writer per_genome_domain_promiscuity_statistics_writer,
1619                                                           final GenomeWideCombinableDomains gwcd,
1620                                                           final int i,
1621                                                           final GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1622         File dc_outfile = new File( input_file_properties[ i ][ 1 ]
1623                 + surfacing.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
1624         if ( output_dir != null ) {
1625             dc_outfile = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile );
1626         }
1627         checkForOutputFileWriteability( dc_outfile );
1628         try {
1629             final BufferedWriter out = new BufferedWriter( new FileWriter( dc_outfile ) );
1630             out.write( gwcd.toStringBuilder( dc_sort_order ).toString() );
1631             out.close();
1632         }
1633         catch ( final IOException e ) {
1634             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1635         }
1636         final DescriptiveStatistics stats = gwcd.getPerGenomeDomainPromiscuityStatistics();
1637         try {
1638             per_genome_domain_promiscuity_statistics_writer.write( input_file_properties[ i ][ 1 ] + "\t" );
1639             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.arithmeticMean() ) + "\t" );
1640             if ( stats.getN() < 2 ) {
1641                 per_genome_domain_promiscuity_statistics_writer.write( "n/a" + "\t" );
1642             }
1643             else {
1644                 per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats
1645                         .sampleStandardDeviation() ) + "\t" );
1646             }
1647             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.median() ) + "\t" );
1648             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMin() + "\t" );
1649             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMax() + "\t" );
1650             per_genome_domain_promiscuity_statistics_writer.write( stats.getN() + "\t" );
1651             final SortedSet<DomainId> mpds = gwcd.getMostPromiscuosDomain();
1652             for( final DomainId mpd : mpds ) {
1653                 per_genome_domain_promiscuity_statistics_writer.write( mpd.getId() + " " );
1654             }
1655             per_genome_domain_promiscuity_statistics_writer.write( ForesterUtil.LINE_SEPARATOR );
1656         }
1657         catch ( final IOException e ) {
1658             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1659         }
1660         if ( input_file_properties[ i ].length == 3 ) {
1661             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1662                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1663                     + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile + "\"" );
1664         }
1665         else {
1666             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1667                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ") to: \""
1668                     + dc_outfile + "\"" );
1669         }
1670     }
1671
1672     public static DescriptiveStatistics writeDomainSimilaritiesToFile( final StringBuilder html_desc,
1673                                                                        final StringBuilder html_title,
1674                                                                        final Writer single_writer,
1675                                                                        Map<Character, Writer> split_writers,
1676                                                                        final SortedSet<DomainSimilarity> similarities,
1677                                                                        final boolean treat_as_binary,
1678                                                                        final List<Species> species_order,
1679                                                                        final PrintableDomainSimilarity.PRINT_OPTION print_option,
1680                                                                        final DomainSimilarity.DomainSimilaritySortField sort_field,
1681                                                                        final DomainSimilarity.DomainSimilarityScoring scoring,
1682                                                                        final boolean verbose,
1683                                                                        final Map<String, Integer> tax_code_to_id_map )
1684             throws IOException {
1685         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
1686         String histogram_title = null;
1687         switch ( sort_field ) {
1688             case ABS_MAX_COUNTS_DIFFERENCE:
1689                 if ( treat_as_binary ) {
1690                     histogram_title = "absolute counts difference:";
1691                 }
1692                 else {
1693                     histogram_title = "absolute (maximal) counts difference:";
1694                 }
1695                 break;
1696             case MAX_COUNTS_DIFFERENCE:
1697                 if ( treat_as_binary ) {
1698                     histogram_title = "counts difference:";
1699                 }
1700                 else {
1701                     histogram_title = "(maximal) counts difference:";
1702                 }
1703                 break;
1704             case DOMAIN_ID:
1705                 histogram_title = "score mean:";
1706                 break;
1707             case MIN:
1708                 histogram_title = "score minimum:";
1709                 break;
1710             case MAX:
1711                 histogram_title = "score maximum:";
1712                 break;
1713             case MAX_DIFFERENCE:
1714                 if ( treat_as_binary ) {
1715                     histogram_title = "difference:";
1716                 }
1717                 else {
1718                     histogram_title = "(maximal) difference:";
1719                 }
1720                 break;
1721             case MEAN:
1722                 histogram_title = "score mean:";
1723                 break;
1724             case SD:
1725                 histogram_title = "score standard deviation:";
1726                 break;
1727             case SPECIES_COUNT:
1728                 histogram_title = "species number:";
1729                 break;
1730             default:
1731                 throw new AssertionError( "Unknown sort field: " + sort_field );
1732         }
1733         for( final DomainSimilarity similarity : similarities ) {
1734             switch ( sort_field ) {
1735                 case ABS_MAX_COUNTS_DIFFERENCE:
1736                     stats.addValue( Math.abs( similarity.getMaximalDifferenceInCounts() ) );
1737                     break;
1738                 case MAX_COUNTS_DIFFERENCE:
1739                     stats.addValue( similarity.getMaximalDifferenceInCounts() );
1740                     break;
1741                 case DOMAIN_ID:
1742                     stats.addValue( similarity.getMeanSimilarityScore() );
1743                     break;
1744                 case MIN:
1745                     stats.addValue( similarity.getMinimalSimilarityScore() );
1746                     break;
1747                 case MAX:
1748                     stats.addValue( similarity.getMaximalSimilarityScore() );
1749                     break;
1750                 case MAX_DIFFERENCE:
1751                     stats.addValue( similarity.getMaximalDifference() );
1752                     break;
1753                 case MEAN:
1754                     stats.addValue( similarity.getMeanSimilarityScore() );
1755                     break;
1756                 case SD:
1757                     stats.addValue( similarity.getStandardDeviationOfSimilarityScore() );
1758                     break;
1759                 case SPECIES_COUNT:
1760                     stats.addValue( similarity.getSpecies().size() );
1761                     break;
1762                 default:
1763                     throw new AssertionError( "Unknown sort field: " + sort_field );
1764             }
1765         }
1766         AsciiHistogram histo = null;
1767         if ( stats.getMin() < stats.getMin() ) {
1768             histo = new AsciiHistogram( stats, histogram_title );
1769         }
1770         if ( verbose ) {
1771             if ( histo != null ) {
1772                 System.out.println( histo.toStringBuffer( 20, '|', 40, 5 ) );
1773             }
1774             System.out.println();
1775             System.out.println( "N                   : " + stats.getN() );
1776             System.out.println( "Min                 : " + stats.getMin() );
1777             System.out.println( "Max                 : " + stats.getMax() );
1778             System.out.println( "Mean                : " + stats.arithmeticMean() );
1779             if ( stats.getN() > 1 ) {
1780                 System.out.println( "SD                  : " + stats.sampleStandardDeviation() );
1781             }
1782             else {
1783                 System.out.println( "SD                  : n/a" );
1784             }
1785             System.out.println( "Median              : " + stats.median() );
1786             if ( stats.getN() > 1 ) {
1787                 System.out.println( "Pearsonian skewness : " + stats.pearsonianSkewness() );
1788             }
1789             else {
1790                 System.out.println( "Pearsonian skewness : n/a" );
1791             }
1792         }
1793         if ( ( single_writer != null ) && ( ( split_writers == null ) || split_writers.isEmpty() ) ) {
1794             split_writers = new HashMap<Character, Writer>();
1795             split_writers.put( '_', single_writer );
1796         }
1797         switch ( print_option ) {
1798             case SIMPLE_TAB_DELIMITED:
1799                 break;
1800             case HTML:
1801                 for( final Character key : split_writers.keySet() ) {
1802                     final Writer w = split_writers.get( key );
1803                     w.write( "<html>" );
1804                     w.write( SurfacingConstants.NL );
1805                     if ( key != '_' ) {
1806                         addHtmlHead( w, "DCs (" + html_title + ") " + key.toString().toUpperCase() );
1807                     }
1808                     else {
1809                         addHtmlHead( w, "DCs (" + html_title + ")" );
1810                     }
1811                     w.write( SurfacingConstants.NL );
1812                     w.write( "<body>" );
1813                     w.write( SurfacingConstants.NL );
1814                     w.write( html_desc.toString() );
1815                     w.write( SurfacingConstants.NL );
1816                     w.write( "<hr>" );
1817                     w.write( "<br>" );
1818                     w.write( SurfacingConstants.NL );
1819                     w.write( "<tt><pre>" );
1820                     w.write( SurfacingConstants.NL );
1821                     if ( histo != null ) {
1822                         w.write( histo.toStringBuffer( 20, '|', 40, 5 ).toString() );
1823                         w.write( SurfacingConstants.NL );
1824                     }
1825                     w.write( "</pre></tt>" );
1826                     w.write( SurfacingConstants.NL );
1827                     w.write( "<table>" );
1828                     w.write( SurfacingConstants.NL );
1829                     w.write( "<tr><td>N: </td><td>" + stats.getN() + "</td></tr>" );
1830                     w.write( SurfacingConstants.NL );
1831                     w.write( "<tr><td>Min: </td><td>" + stats.getMin() + "</td></tr>" );
1832                     w.write( SurfacingConstants.NL );
1833                     w.write( "<tr><td>Max: </td><td>" + stats.getMax() + "</td></tr>" );
1834                     w.write( SurfacingConstants.NL );
1835                     w.write( "<tr><td>Mean: </td><td>" + stats.arithmeticMean() + "</td></tr>" );
1836                     w.write( SurfacingConstants.NL );
1837                     if ( stats.getN() > 1 ) {
1838                         w.write( "<tr><td>SD: </td><td>" + stats.sampleStandardDeviation() + "</td></tr>" );
1839                     }
1840                     else {
1841                         w.write( "<tr><td>SD: </td><td>n/a</td></tr>" );
1842                     }
1843                     w.write( SurfacingConstants.NL );
1844                     w.write( "<tr><td>Median: </td><td>" + stats.median() + "</td></tr>" );
1845                     w.write( SurfacingConstants.NL );
1846                     if ( stats.getN() > 1 ) {
1847                         w.write( "<tr><td>Pearsonian skewness: </td><td>" + stats.pearsonianSkewness() + "</td></tr>" );
1848                     }
1849                     else {
1850                         w.write( "<tr><td>Pearsonian skewness: </td><td>n/a</td></tr>" );
1851                     }
1852                     w.write( SurfacingConstants.NL );
1853                     w.write( "</table>" );
1854                     w.write( SurfacingConstants.NL );
1855                     w.write( "<br>" );
1856                     w.write( SurfacingConstants.NL );
1857                     w.write( "<hr>" );
1858                     w.write( SurfacingConstants.NL );
1859                     w.write( "<br>" );
1860                     w.write( SurfacingConstants.NL );
1861                     w.write( "<table>" );
1862                     w.write( SurfacingConstants.NL );
1863                 }
1864                 break;
1865         }
1866         for( final Writer w : split_writers.values() ) {
1867             w.write( SurfacingConstants.NL );
1868         }
1869         for( final DomainSimilarity similarity : similarities ) {
1870             if ( ( species_order != null ) && !species_order.isEmpty() ) {
1871                 ( ( PrintableDomainSimilarity ) similarity ).setSpeciesOrder( species_order );
1872             }
1873             if ( single_writer != null ) {
1874                 single_writer.write( similarity.toStringBuffer( print_option, tax_code_to_id_map ).toString() );
1875             }
1876             else {
1877                 Writer local_writer = split_writers.get( ( similarity.getDomainId().getId().charAt( 0 ) + "" )
1878                         .toLowerCase().charAt( 0 ) );
1879                 if ( local_writer == null ) {
1880                     local_writer = split_writers.get( '0' );
1881                 }
1882                 local_writer.write( similarity.toStringBuffer( print_option, tax_code_to_id_map ).toString() );
1883             }
1884             for( final Writer w : split_writers.values() ) {
1885                 w.write( SurfacingConstants.NL );
1886             }
1887         }
1888         switch ( print_option ) {
1889             case HTML:
1890                 for( final Writer w : split_writers.values() ) {
1891                     w.write( SurfacingConstants.NL );
1892                     w.write( "</table>" );
1893                     w.write( SurfacingConstants.NL );
1894                     w.write( "</font>" );
1895                     w.write( SurfacingConstants.NL );
1896                     w.write( "</body>" );
1897                     w.write( SurfacingConstants.NL );
1898                     w.write( "</html>" );
1899                     w.write( SurfacingConstants.NL );
1900                 }
1901                 break;
1902         }
1903         for( final Writer w : split_writers.values() ) {
1904             w.close();
1905         }
1906         return stats;
1907     }
1908
1909     public static void writeMatrixToFile( final CharacterStateMatrix<?> matrix,
1910                                           final String filename,
1911                                           final Format format ) {
1912         final File outfile = new File( filename );
1913         checkForOutputFileWriteability( outfile );
1914         try {
1915             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1916             matrix.toWriter( out, format );
1917             out.flush();
1918             out.close();
1919         }
1920         catch ( final IOException e ) {
1921             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1922         }
1923         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote matrix: \"" + filename + "\"" );
1924     }
1925
1926     public static void writeMatrixToFile( final File matrix_outfile, final List<DistanceMatrix> matrices ) {
1927         checkForOutputFileWriteability( matrix_outfile );
1928         try {
1929             final BufferedWriter out = new BufferedWriter( new FileWriter( matrix_outfile ) );
1930             for( final DistanceMatrix distance_matrix : matrices ) {
1931                 out.write( distance_matrix.toStringBuffer( DistanceMatrix.Format.PHYLIP ).toString() );
1932                 out.write( ForesterUtil.LINE_SEPARATOR );
1933                 out.flush();
1934             }
1935             out.close();
1936         }
1937         catch ( final IOException e ) {
1938             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1939         }
1940         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote distance matrices to \"" + matrix_outfile + "\"" );
1941     }
1942
1943     public static void writePhylogenyToFile( final Phylogeny phylogeny, final String filename ) {
1944         final PhylogenyWriter writer = new PhylogenyWriter();
1945         try {
1946             writer.toPhyloXML( new File( filename ), phylogeny, 1 );
1947         }
1948         catch ( final IOException e ) {
1949             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "failed to write phylogeny to \"" + filename + "\": "
1950                     + e );
1951         }
1952         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote phylogeny to \"" + filename + "\"" );
1953     }
1954
1955     public static void writeTaxonomyLinks( final Writer writer,
1956                                            final String species,
1957                                            final Map<String, Integer> tax_code_to_id_map ) throws IOException {
1958         if ( ( species.length() > 1 ) && ( species.indexOf( '_' ) < 1 ) ) {
1959             writer.write( " [" );
1960             if ( ( tax_code_to_id_map != null ) && tax_code_to_id_map.containsKey( species ) ) {
1961                 writer.write( "<a href=\"" + SurfacingConstants.UNIPROT_TAXONOMY_ID_LINK
1962                         + tax_code_to_id_map.get( species ) + "\" target=\"taxonomy_window\">uniprot</a>" );
1963             }
1964             else {
1965                 writer.write( "<a href=\"" + SurfacingConstants.EOL_LINK + species
1966                         + "\" target=\"taxonomy_window\">eol</a>" );
1967                 writer.write( "|" );
1968                 writer.write( "<a href=\"" + SurfacingConstants.GOOGLE_SCHOLAR_SEARCH + species
1969                         + "\" target=\"taxonomy_window\">scholar</a>" );
1970                 writer.write( "|" );
1971                 writer.write( "<a href=\"" + SurfacingConstants.GOOGLE_WEB_SEARCH_LINK + species
1972                         + "\" target=\"taxonomy_window\">google</a>" );
1973             }
1974             writer.write( "]" );
1975         }
1976     }
1977
1978     private final static void addToCountMap( final Map<String, Integer> map, final String s ) {
1979         if ( map.containsKey( s ) ) {
1980             map.put( s, map.get( s ) + 1 );
1981         }
1982         else {
1983             map.put( s, 1 );
1984         }
1985     }
1986
1987     private static void calculateIndependentDomainCombinationGains( final Phylogeny local_phylogeny_l,
1988                                                                     final String outfilename_for_counts,
1989                                                                     final String outfilename_for_dc,
1990                                                                     final String outfilename_for_dc_for_go_mapping,
1991                                                                     final String outfilename_for_dc_for_go_mapping_unique,
1992                                                                     final String outfilename_for_rank_counts,
1993                                                                     final String outfilename_for_ancestor_species_counts,
1994                                                                     final String outfilename_for_protein_stats,
1995                                                                     final Map<String, DescriptiveStatistics> protein_length_stats_by_dc,
1996                                                                     final Map<String, DescriptiveStatistics> domain_number_stats_by_dc,
1997                                                                     final Map<String, DescriptiveStatistics> domain_length_stats_by_domain ) {
1998         try {
1999             //
2000             //            if ( protein_length_stats_by_dc != null ) {
2001             //                for( final Entry<?, DescriptiveStatistics> entry : protein_length_stats_by_dc.entrySet() ) {
2002             //                    System.out.print( entry.getKey().toString() );
2003             //                    System.out.print( ": " );
2004             //                    double[] a = entry.getValue().getDataAsDoubleArray();
2005             //                    for( int i = 0; i < a.length; i++ ) {
2006             //                        System.out.print( a[ i ] + " " );
2007             //                    }
2008             //                    System.out.println();
2009             //                }
2010             //            }
2011             //            if ( domain_number_stats_by_dc != null ) {
2012             //                for( final Entry<?, DescriptiveStatistics> entry : domain_number_stats_by_dc.entrySet() ) {
2013             //                    System.out.print( entry.getKey().toString() );
2014             //                    System.out.print( ": " );
2015             //                    double[] a = entry.getValue().getDataAsDoubleArray();
2016             //                    for( int i = 0; i < a.length; i++ ) {
2017             //                        System.out.print( a[ i ] + " " );
2018             //                    }
2019             //                    System.out.println();
2020             //                }
2021             //            }
2022             //
2023             final BufferedWriter out_counts = new BufferedWriter( new FileWriter( outfilename_for_counts ) );
2024             final BufferedWriter out_dc = new BufferedWriter( new FileWriter( outfilename_for_dc ) );
2025             final BufferedWriter out_dc_for_go_mapping = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping ) );
2026             final BufferedWriter out_dc_for_go_mapping_unique = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping_unique ) );
2027             final SortedMap<String, Integer> dc_gain_counts = new TreeMap<String, Integer>();
2028             for( final PhylogenyNodeIterator it = local_phylogeny_l.iteratorPostorder(); it.hasNext(); ) {
2029                 final PhylogenyNode n = it.next();
2030                 final Set<String> gained_dc = n.getNodeData().getBinaryCharacters().getGainedCharacters();
2031                 for( final String dc : gained_dc ) {
2032                     if ( dc_gain_counts.containsKey( dc ) ) {
2033                         dc_gain_counts.put( dc, dc_gain_counts.get( dc ) + 1 );
2034                     }
2035                     else {
2036                         dc_gain_counts.put( dc, 1 );
2037                     }
2038                 }
2039             }
2040             final SortedMap<Integer, Integer> histogram = new TreeMap<Integer, Integer>();
2041             final SortedMap<Integer, StringBuilder> domain_lists = new TreeMap<Integer, StringBuilder>();
2042             final SortedMap<Integer, DescriptiveStatistics> dc_reapp_counts_to_protein_length_stats = new TreeMap<Integer, DescriptiveStatistics>();
2043             final SortedMap<Integer, DescriptiveStatistics> dc_reapp_counts_to_domain_number_stats = new TreeMap<Integer, DescriptiveStatistics>();
2044             final SortedMap<Integer, DescriptiveStatistics> dc_reapp_counts_to_domain_lengths_stats = new TreeMap<Integer, DescriptiveStatistics>();
2045             final SortedMap<Integer, PriorityQueue<String>> domain_lists_go = new TreeMap<Integer, PriorityQueue<String>>();
2046             final SortedMap<Integer, SortedSet<String>> domain_lists_go_unique = new TreeMap<Integer, SortedSet<String>>();
2047             final Set<String> dcs = dc_gain_counts.keySet();
2048             final SortedSet<String> more_than_once = new TreeSet<String>();
2049             final DescriptiveStatistics gained_once_lengths_stats = new BasicDescriptiveStatistics();
2050             final DescriptiveStatistics gained_once_domain_count_stats = new BasicDescriptiveStatistics();
2051             final DescriptiveStatistics gained_multiple_times_lengths_stats = new BasicDescriptiveStatistics();
2052             final DescriptiveStatistics gained_multiple_times_domain_count_stats = new BasicDescriptiveStatistics();
2053             long gained_multiple_times_domain_length_sum = 0;
2054             long gained_once_domain_length_sum = 0;
2055             long gained_multiple_times_domain_length_count = 0;
2056             long gained_once_domain_length_count = 0;
2057             for( final String dc : dcs ) {
2058                 final int count = dc_gain_counts.get( dc );
2059                 if ( histogram.containsKey( count ) ) {
2060                     histogram.put( count, histogram.get( count ) + 1 );
2061                     domain_lists.get( count ).append( ", " + dc );
2062                     domain_lists_go.get( count ).addAll( splitDomainCombination( dc ) );
2063                     domain_lists_go_unique.get( count ).addAll( splitDomainCombination( dc ) );
2064                 }
2065                 else {
2066                     histogram.put( count, 1 );
2067                     domain_lists.put( count, new StringBuilder( dc ) );
2068                     final PriorityQueue<String> q = new PriorityQueue<String>();
2069                     q.addAll( splitDomainCombination( dc ) );
2070                     domain_lists_go.put( count, q );
2071                     final SortedSet<String> set = new TreeSet<String>();
2072                     set.addAll( splitDomainCombination( dc ) );
2073                     domain_lists_go_unique.put( count, set );
2074                 }
2075                 if ( protein_length_stats_by_dc != null ) {
2076                     if ( !dc_reapp_counts_to_protein_length_stats.containsKey( count ) ) {
2077                         dc_reapp_counts_to_protein_length_stats.put( count, new BasicDescriptiveStatistics() );
2078                     }
2079                     dc_reapp_counts_to_protein_length_stats.get( count ).addValue( protein_length_stats_by_dc.get( dc )
2080                             .arithmeticMean() );
2081                 }
2082                 if ( domain_number_stats_by_dc != null ) {
2083                     if ( !dc_reapp_counts_to_domain_number_stats.containsKey( count ) ) {
2084                         dc_reapp_counts_to_domain_number_stats.put( count, new BasicDescriptiveStatistics() );
2085                     }
2086                     dc_reapp_counts_to_domain_number_stats.get( count ).addValue( domain_number_stats_by_dc.get( dc )
2087                             .arithmeticMean() );
2088                 }
2089                 if ( domain_length_stats_by_domain != null ) {
2090                     if ( !dc_reapp_counts_to_domain_lengths_stats.containsKey( count ) ) {
2091                         dc_reapp_counts_to_domain_lengths_stats.put( count, new BasicDescriptiveStatistics() );
2092                     }
2093                     final String[] ds = dc.split( "=" );
2094                     dc_reapp_counts_to_domain_lengths_stats.get( count ).addValue( domain_length_stats_by_domain
2095                             .get( ds[ 0 ] ).arithmeticMean() );
2096                     dc_reapp_counts_to_domain_lengths_stats.get( count ).addValue( domain_length_stats_by_domain
2097                             .get( ds[ 1 ] ).arithmeticMean() );
2098                 }
2099                 if ( count > 1 ) {
2100                     more_than_once.add( dc );
2101                     if ( protein_length_stats_by_dc != null ) {
2102                         final DescriptiveStatistics s = protein_length_stats_by_dc.get( dc );
2103                         for( final double element : s.getData() ) {
2104                             gained_multiple_times_lengths_stats.addValue( element );
2105                         }
2106                     }
2107                     if ( domain_number_stats_by_dc != null ) {
2108                         final DescriptiveStatistics s = domain_number_stats_by_dc.get( dc );
2109                         for( final double element : s.getData() ) {
2110                             gained_multiple_times_domain_count_stats.addValue( element );
2111                         }
2112                     }
2113                     if ( domain_length_stats_by_domain != null ) {
2114                         final String[] ds = dc.split( "=" );
2115                         final DescriptiveStatistics s0 = domain_length_stats_by_domain.get( ds[ 0 ] );
2116                         final DescriptiveStatistics s1 = domain_length_stats_by_domain.get( ds[ 1 ] );
2117                         for( final double element : s0.getData() ) {
2118                             gained_multiple_times_domain_length_sum += element;
2119                             ++gained_multiple_times_domain_length_count;
2120                         }
2121                         for( final double element : s1.getData() ) {
2122                             gained_multiple_times_domain_length_sum += element;
2123                             ++gained_multiple_times_domain_length_count;
2124                         }
2125                     }
2126                 }
2127                 else {
2128                     if ( protein_length_stats_by_dc != null ) {
2129                         final DescriptiveStatistics s = protein_length_stats_by_dc.get( dc );
2130                         for( final double element : s.getData() ) {
2131                             gained_once_lengths_stats.addValue( element );
2132                         }
2133                     }
2134                     if ( domain_number_stats_by_dc != null ) {
2135                         final DescriptiveStatistics s = domain_number_stats_by_dc.get( dc );
2136                         for( final double element : s.getData() ) {
2137                             gained_once_domain_count_stats.addValue( element );
2138                         }
2139                     }
2140                     if ( domain_length_stats_by_domain != null ) {
2141                         final String[] ds = dc.split( "=" );
2142                         final DescriptiveStatistics s0 = domain_length_stats_by_domain.get( ds[ 0 ] );
2143                         final DescriptiveStatistics s1 = domain_length_stats_by_domain.get( ds[ 1 ] );
2144                         for( final double element : s0.getData() ) {
2145                             gained_once_domain_length_sum += element;
2146                             ++gained_once_domain_length_count;
2147                         }
2148                         for( final double element : s1.getData() ) {
2149                             gained_once_domain_length_sum += element;
2150                             ++gained_once_domain_length_count;
2151                         }
2152                     }
2153                 }
2154             }
2155             final Set<Integer> histogram_keys = histogram.keySet();
2156             for( final Integer histogram_key : histogram_keys ) {
2157                 final int count = histogram.get( histogram_key );
2158                 final StringBuilder dc = domain_lists.get( histogram_key );
2159                 out_counts.write( histogram_key + "\t" + count + ForesterUtil.LINE_SEPARATOR );
2160                 out_dc.write( histogram_key + "\t" + dc + ForesterUtil.LINE_SEPARATOR );
2161                 out_dc_for_go_mapping.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
2162                 final Object[] sorted = domain_lists_go.get( histogram_key ).toArray();
2163                 Arrays.sort( sorted );
2164                 for( final Object domain : sorted ) {
2165                     out_dc_for_go_mapping.write( domain + ForesterUtil.LINE_SEPARATOR );
2166                 }
2167                 out_dc_for_go_mapping_unique.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
2168                 for( final String domain : domain_lists_go_unique.get( histogram_key ) ) {
2169                     out_dc_for_go_mapping_unique.write( domain + ForesterUtil.LINE_SEPARATOR );
2170                 }
2171             }
2172             out_counts.close();
2173             out_dc.close();
2174             out_dc_for_go_mapping.close();
2175             out_dc_for_go_mapping_unique.close();
2176             final SortedMap<String, Integer> lca_rank_counts = new TreeMap<String, Integer>();
2177             final SortedMap<String, Integer> lca_ancestor_species_counts = new TreeMap<String, Integer>();
2178             for( final String dc : more_than_once ) {
2179                 final List<PhylogenyNode> nodes = new ArrayList<PhylogenyNode>();
2180                 for( final PhylogenyNodeIterator it = local_phylogeny_l.iteratorExternalForward(); it.hasNext(); ) {
2181                     final PhylogenyNode n = it.next();
2182                     if ( n.getNodeData().getBinaryCharacters().getGainedCharacters().contains( dc ) ) {
2183                         nodes.add( n );
2184                     }
2185                 }
2186                 for( int i = 0; i < ( nodes.size() - 1 ); ++i ) {
2187                     for( int j = i + 1; j < nodes.size(); ++j ) {
2188                         final PhylogenyNode lca = PhylogenyMethods.calculateLCA( nodes.get( i ), nodes.get( j ) );
2189                         String rank = "unknown";
2190                         if ( lca.getNodeData().isHasTaxonomy()
2191                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getRank() ) ) {
2192                             rank = lca.getNodeData().getTaxonomy().getRank();
2193                         }
2194                         addToCountMap( lca_rank_counts, rank );
2195                         String lca_species;
2196                         if ( lca.getNodeData().isHasTaxonomy()
2197                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getScientificName() ) ) {
2198                             lca_species = lca.getNodeData().getTaxonomy().getScientificName();
2199                         }
2200                         else if ( lca.getNodeData().isHasTaxonomy()
2201                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getCommonName() ) ) {
2202                             lca_species = lca.getNodeData().getTaxonomy().getCommonName();
2203                         }
2204                         else {
2205                             lca_species = lca.getName();
2206                         }
2207                         addToCountMap( lca_ancestor_species_counts, lca_species );
2208                     }
2209                 }
2210             }
2211             final BufferedWriter out_for_rank_counts = new BufferedWriter( new FileWriter( outfilename_for_rank_counts ) );
2212             final BufferedWriter out_for_ancestor_species_counts = new BufferedWriter( new FileWriter( outfilename_for_ancestor_species_counts ) );
2213             ForesterUtil.map2writer( out_for_rank_counts, lca_rank_counts, "\t", ForesterUtil.LINE_SEPARATOR );
2214             ForesterUtil.map2writer( out_for_ancestor_species_counts,
2215                                      lca_ancestor_species_counts,
2216                                      "\t",
2217                                      ForesterUtil.LINE_SEPARATOR );
2218             out_for_rank_counts.close();
2219             out_for_ancestor_species_counts.close();
2220             if ( !ForesterUtil.isEmpty( outfilename_for_protein_stats )
2221                     && ( ( domain_length_stats_by_domain != null ) || ( protein_length_stats_by_dc != null ) || ( domain_number_stats_by_dc != null ) ) ) {
2222                 final BufferedWriter w = new BufferedWriter( new FileWriter( outfilename_for_protein_stats ) );
2223                 w.write( "Domain Lengths: " );
2224                 w.write( "\n" );
2225                 if ( domain_length_stats_by_domain != null ) {
2226                     for( final Entry<Integer, DescriptiveStatistics> entry : dc_reapp_counts_to_domain_lengths_stats
2227                             .entrySet() ) {
2228                         w.write( entry.getKey().toString() );
2229                         w.write( "\t" + entry.getValue().arithmeticMean() );
2230                         w.write( "\t" + entry.getValue().median() );
2231                         w.write( "\n" );
2232                     }
2233                 }
2234                 w.flush();
2235                 w.write( "\n" );
2236                 w.write( "\n" );
2237                 w.write( "Protein Lengths: " );
2238                 w.write( "\n" );
2239                 if ( protein_length_stats_by_dc != null ) {
2240                     for( final Entry<Integer, DescriptiveStatistics> entry : dc_reapp_counts_to_protein_length_stats
2241                             .entrySet() ) {
2242                         w.write( entry.getKey().toString() );
2243                         w.write( "\t" + entry.getValue().arithmeticMean() );
2244                         w.write( "\t" + entry.getValue().median() );
2245                         w.write( "\n" );
2246                     }
2247                 }
2248                 w.flush();
2249                 w.write( "\n" );
2250                 w.write( "\n" );
2251                 w.write( "Number of domains: " );
2252                 w.write( "\n" );
2253                 if ( domain_number_stats_by_dc != null ) {
2254                     for( final Entry<Integer, DescriptiveStatistics> entry : dc_reapp_counts_to_domain_number_stats
2255                             .entrySet() ) {
2256                         w.write( entry.getKey().toString() );
2257                         w.write( "\t" + entry.getValue().arithmeticMean() );
2258                         w.write( "\t" + entry.getValue().median() );
2259                         w.write( "\n" );
2260                     }
2261                 }
2262                 w.flush();
2263                 w.write( "\n" );
2264                 w.write( "\n" );
2265                 w.write( "Gained once, domain lengths:" );
2266                 w.write( "\n" );
2267                 w.write( "N: " + gained_once_domain_length_count );
2268                 w.write( "\n" );
2269                 w.write( "Avg: " + ( ( double ) gained_once_domain_length_sum / gained_once_domain_length_count ) );
2270                 w.write( "\n" );
2271                 w.write( "\n" );
2272                 w.write( "Gained multiple times, domain lengths:" );
2273                 w.write( "\n" );
2274                 w.write( "N: " + gained_multiple_times_domain_length_count );
2275                 w.write( "\n" );
2276                 w.write( "Avg: "
2277                         + ( ( double ) gained_multiple_times_domain_length_sum / gained_multiple_times_domain_length_count ) );
2278                 w.write( "\n" );
2279                 w.write( "\n" );
2280                 w.write( "\n" );
2281                 w.write( "\n" );
2282                 w.write( "Gained once, protein lengths:" );
2283                 w.write( "\n" );
2284                 w.write( gained_once_lengths_stats.toString() );
2285                 w.write( "\n" );
2286                 w.write( "\n" );
2287                 w.write( "Gained once, domain counts:" );
2288                 w.write( "\n" );
2289                 w.write( gained_once_domain_count_stats.toString() );
2290                 w.write( "\n" );
2291                 w.write( "\n" );
2292                 w.write( "Gained multiple times, protein lengths:" );
2293                 w.write( "\n" );
2294                 w.write( gained_multiple_times_lengths_stats.toString() );
2295                 w.write( "\n" );
2296                 w.write( "\n" );
2297                 w.write( "Gained multiple times, domain counts:" );
2298                 w.write( "\n" );
2299                 w.write( gained_multiple_times_domain_count_stats.toString() );
2300                 w.flush();
2301                 w.close();
2302             }
2303         }
2304         catch ( final IOException e ) {
2305             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
2306         }
2307         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch counts to ["
2308                 + outfilename_for_counts + "]" );
2309         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch lists to ["
2310                 + outfilename_for_dc + "]" );
2311         ForesterUtil.programMessage( surfacing.PRG_NAME,
2312                                      "Wrote independent domain combination gains fitch lists to (for GO mapping) ["
2313                                              + outfilename_for_dc_for_go_mapping + "]" );
2314         ForesterUtil.programMessage( surfacing.PRG_NAME,
2315                                      "Wrote independent domain combination gains fitch lists to (for GO mapping, unique) ["
2316                                              + outfilename_for_dc_for_go_mapping_unique + "]" );
2317     }
2318
2319     private static SortedSet<String> collectAllDomainsChangedOnSubtree( final PhylogenyNode subtree_root,
2320                                                                         final boolean get_gains ) {
2321         final SortedSet<String> domains = new TreeSet<String>();
2322         for( final PhylogenyNode descendant : PhylogenyMethods.getAllDescendants( subtree_root ) ) {
2323             final BinaryCharacters chars = descendant.getNodeData().getBinaryCharacters();
2324             if ( get_gains ) {
2325                 domains.addAll( chars.getGainedCharacters() );
2326             }
2327             else {
2328                 domains.addAll( chars.getLostCharacters() );
2329             }
2330         }
2331         return domains;
2332     }
2333
2334     private static File createBaseDirForPerNodeDomainFiles( final String base_dir,
2335                                                             final boolean domain_combinations,
2336                                                             final CharacterStateMatrix.GainLossStates state,
2337                                                             final String outfile ) {
2338         File per_node_go_mapped_domain_gain_loss_files_base_dir = new File( new File( outfile ).getParent()
2339                 + ForesterUtil.FILE_SEPARATOR + base_dir );
2340         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
2341             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
2342         }
2343         if ( domain_combinations ) {
2344             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
2345                     + ForesterUtil.FILE_SEPARATOR + "DC" );
2346         }
2347         else {
2348             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
2349                     + ForesterUtil.FILE_SEPARATOR + "DOMAINS" );
2350         }
2351         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
2352             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
2353         }
2354         if ( state == GainLossStates.GAIN ) {
2355             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
2356                     + ForesterUtil.FILE_SEPARATOR + "GAINS" );
2357         }
2358         else if ( state == GainLossStates.LOSS ) {
2359             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
2360                     + ForesterUtil.FILE_SEPARATOR + "LOSSES" );
2361         }
2362         else {
2363             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
2364                     + ForesterUtil.FILE_SEPARATOR + "PRESENT" );
2365         }
2366         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
2367             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
2368         }
2369         return per_node_go_mapped_domain_gain_loss_files_base_dir;
2370     }
2371
2372     private static SortedSet<BinaryDomainCombination> createSetOfAllBinaryDomainCombinationsPerGenome( final GenomeWideCombinableDomains gwcd ) {
2373         final SortedMap<DomainId, CombinableDomains> cds = gwcd.getAllCombinableDomainsIds();
2374         final SortedSet<BinaryDomainCombination> binary_combinations = new TreeSet<BinaryDomainCombination>();
2375         for( final DomainId domain_id : cds.keySet() ) {
2376             final CombinableDomains cd = cds.get( domain_id );
2377             binary_combinations.addAll( cd.toBinaryDomainCombinations() );
2378         }
2379         return binary_combinations;
2380     }
2381
2382     private static List<String> splitDomainCombination( final String dc ) {
2383         final String[] s = dc.split( "=" );
2384         if ( s.length != 2 ) {
2385             ForesterUtil.printErrorMessage( surfacing.PRG_NAME, "Stringyfied domain combination has illegal format: "
2386                     + dc );
2387             System.exit( -1 );
2388         }
2389         final List<String> l = new ArrayList<String>( 2 );
2390         l.add( s[ 0 ] );
2391         l.add( s[ 1 ] );
2392         return l;
2393     }
2394
2395     private static void writeAllEncounteredPfamsToFile( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
2396                                                         final Map<GoId, GoTerm> go_id_to_term_map,
2397                                                         final String outfile_name,
2398                                                         final SortedSet<String> all_pfams_encountered ) {
2399         final File all_pfams_encountered_file = new File( outfile_name + surfacing.ALL_PFAMS_ENCOUNTERED_SUFFIX );
2400         final File all_pfams_encountered_with_go_annotation_file = new File( outfile_name
2401                 + surfacing.ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX );
2402         final File encountered_pfams_summary_file = new File( outfile_name + surfacing.ENCOUNTERED_PFAMS_SUMMARY_SUFFIX );
2403         int biological_process_counter = 0;
2404         int cellular_component_counter = 0;
2405         int molecular_function_counter = 0;
2406         int pfams_with_mappings_counter = 0;
2407         int pfams_without_mappings_counter = 0;
2408         int pfams_without_mappings_to_bp_or_mf_counter = 0;
2409         int pfams_with_mappings_to_bp_or_mf_counter = 0;
2410         try {
2411             final Writer all_pfams_encountered_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_file ) );
2412             final Writer all_pfams_encountered_with_go_annotation_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_with_go_annotation_file ) );
2413             final Writer summary_writer = new BufferedWriter( new FileWriter( encountered_pfams_summary_file ) );
2414             summary_writer.write( "# Pfam to GO mapping summary" );
2415             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2416             summary_writer.write( "# Actual summary is at the end of this file." );
2417             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2418             summary_writer.write( "# Encountered Pfams without a GO mapping:" );
2419             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2420             for( final String pfam : all_pfams_encountered ) {
2421                 all_pfams_encountered_writer.write( pfam );
2422                 all_pfams_encountered_writer.write( ForesterUtil.LINE_SEPARATOR );
2423                 final DomainId domain_id = new DomainId( pfam );
2424                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
2425                     ++pfams_with_mappings_counter;
2426                     all_pfams_encountered_with_go_annotation_writer.write( pfam );
2427                     all_pfams_encountered_with_go_annotation_writer.write( ForesterUtil.LINE_SEPARATOR );
2428                     final List<GoId> go_ids = domain_id_to_go_ids_map.get( domain_id );
2429                     boolean maps_to_bp = false;
2430                     boolean maps_to_cc = false;
2431                     boolean maps_to_mf = false;
2432                     for( final GoId go_id : go_ids ) {
2433                         final GoTerm go_term = go_id_to_term_map.get( go_id );
2434                         if ( go_term.getGoNameSpace().isBiologicalProcess() ) {
2435                             maps_to_bp = true;
2436                         }
2437                         else if ( go_term.getGoNameSpace().isCellularComponent() ) {
2438                             maps_to_cc = true;
2439                         }
2440                         else if ( go_term.getGoNameSpace().isMolecularFunction() ) {
2441                             maps_to_mf = true;
2442                         }
2443                     }
2444                     if ( maps_to_bp ) {
2445                         ++biological_process_counter;
2446                     }
2447                     if ( maps_to_cc ) {
2448                         ++cellular_component_counter;
2449                     }
2450                     if ( maps_to_mf ) {
2451                         ++molecular_function_counter;
2452                     }
2453                     if ( maps_to_bp || maps_to_mf ) {
2454                         ++pfams_with_mappings_to_bp_or_mf_counter;
2455                     }
2456                     else {
2457                         ++pfams_without_mappings_to_bp_or_mf_counter;
2458                     }
2459                 }
2460                 else {
2461                     ++pfams_without_mappings_to_bp_or_mf_counter;
2462                     ++pfams_without_mappings_counter;
2463                     summary_writer.write( pfam );
2464                     summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2465                 }
2466             }
2467             all_pfams_encountered_writer.close();
2468             all_pfams_encountered_with_go_annotation_writer.close();
2469             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + all_pfams_encountered.size()
2470                     + "] encountered Pfams to: \"" + all_pfams_encountered_file + "\"" );
2471             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + pfams_with_mappings_counter
2472                     + "] encountered Pfams with GO mappings to: \"" + all_pfams_encountered_with_go_annotation_file
2473                     + "\"" );
2474             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote summary (including all ["
2475                     + pfams_without_mappings_counter + "] encountered Pfams without GO mappings) to: \""
2476                     + encountered_pfams_summary_file + "\"" );
2477             ForesterUtil.programMessage( surfacing.PRG_NAME, "Sum of Pfams encountered                : "
2478                     + all_pfams_encountered.size() );
2479             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without a mapping                 : "
2480                     + pfams_without_mappings_counter + " ["
2481                     + ( ( 100 * pfams_without_mappings_counter ) / all_pfams_encountered.size() ) + "%]" );
2482             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without mapping to proc. or func. : "
2483                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
2484                     + ( ( 100 * pfams_without_mappings_to_bp_or_mf_counter ) / all_pfams_encountered.size() ) + "%]" );
2485             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with a mapping                    : "
2486                     + pfams_with_mappings_counter + " ["
2487                     + ( ( 100 * pfams_with_mappings_counter ) / all_pfams_encountered.size() ) + "%]" );
2488             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with a mapping to proc. or func.  : "
2489                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
2490                     + ( ( 100 * pfams_with_mappings_to_bp_or_mf_counter ) / all_pfams_encountered.size() ) + "%]" );
2491             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with mapping to biological process: "
2492                     + biological_process_counter + " ["
2493                     + ( ( 100 * biological_process_counter ) / all_pfams_encountered.size() ) + "%]" );
2494             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with mapping to molecular function: "
2495                     + molecular_function_counter + " ["
2496                     + ( ( 100 * molecular_function_counter ) / all_pfams_encountered.size() ) + "%]" );
2497             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with mapping to cellular component: "
2498                     + cellular_component_counter + " ["
2499                     + ( ( 100 * cellular_component_counter ) / all_pfams_encountered.size() ) + "%]" );
2500             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2501             summary_writer.write( "# Sum of Pfams encountered                : " + all_pfams_encountered.size() );
2502             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2503             summary_writer.write( "# Pfams without a mapping                 : " + pfams_without_mappings_counter
2504                     + " [" + ( ( 100 * pfams_without_mappings_counter ) / all_pfams_encountered.size() ) + "%]" );
2505             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2506             summary_writer.write( "# Pfams without mapping to proc. or func. : "
2507                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
2508                     + ( ( 100 * pfams_without_mappings_to_bp_or_mf_counter ) / all_pfams_encountered.size() ) + "%]" );
2509             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2510             summary_writer.write( "# Pfams with a mapping                    : " + pfams_with_mappings_counter + " ["
2511                     + ( ( 100 * pfams_with_mappings_counter ) / all_pfams_encountered.size() ) + "%]" );
2512             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2513             summary_writer.write( "# Pfams with a mapping to proc. or func.  : "
2514                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
2515                     + ( ( 100 * pfams_with_mappings_to_bp_or_mf_counter ) / all_pfams_encountered.size() ) + "%]" );
2516             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2517             summary_writer.write( "# Pfams with mapping to biological process: " + biological_process_counter + " ["
2518                     + ( ( 100 * biological_process_counter ) / all_pfams_encountered.size() ) + "%]" );
2519             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2520             summary_writer.write( "# Pfams with mapping to molecular function: " + molecular_function_counter + " ["
2521                     + ( ( 100 * molecular_function_counter ) / all_pfams_encountered.size() ) + "%]" );
2522             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2523             summary_writer.write( "# Pfams with mapping to cellular component: " + cellular_component_counter + " ["
2524                     + ( ( 100 * cellular_component_counter ) / all_pfams_encountered.size() ) + "%]" );
2525             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
2526             summary_writer.close();
2527         }
2528         catch ( final IOException e ) {
2529             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
2530         }
2531     }
2532
2533     private static void writeDomainData( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
2534                                          final Map<GoId, GoTerm> go_id_to_term_map,
2535                                          final GoNameSpace go_namespace_limit,
2536                                          final Writer out,
2537                                          final String domain_0,
2538                                          final String domain_1,
2539                                          final String prefix_for_html,
2540                                          final String character_separator_for_non_html_output,
2541                                          final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
2542                                          final Set<GoId> all_go_ids ) throws IOException {
2543         boolean any_go_annotation_present = false;
2544         boolean first_has_no_go = false;
2545         int domain_count = 2; // To distinguish between domains and binary domain combinations.
2546         if ( ForesterUtil.isEmpty( domain_1 ) ) {
2547             domain_count = 1;
2548         }
2549         // The following has a difficult to understand logic.
2550         for( int d = 0; d < domain_count; ++d ) {
2551             List<GoId> go_ids = null;
2552             boolean go_annotation_present = false;
2553             if ( d == 0 ) {
2554                 final DomainId domain_id = new DomainId( domain_0 );
2555                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
2556                     go_annotation_present = true;
2557                     any_go_annotation_present = true;
2558                     go_ids = domain_id_to_go_ids_map.get( domain_id );
2559                 }
2560                 else {
2561                     first_has_no_go = true;
2562                 }
2563             }
2564             else {
2565                 final DomainId domain_id = new DomainId( domain_1 );
2566                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
2567                     go_annotation_present = true;
2568                     any_go_annotation_present = true;
2569                     go_ids = domain_id_to_go_ids_map.get( domain_id );
2570                 }
2571             }
2572             if ( go_annotation_present ) {
2573                 boolean first = ( ( d == 0 ) || ( ( d == 1 ) && first_has_no_go ) );
2574                 for( final GoId go_id : go_ids ) {
2575                     out.write( "<tr>" );
2576                     if ( first ) {
2577                         first = false;
2578                         writeDomainIdsToHtml( out,
2579                                               domain_0,
2580                                               domain_1,
2581                                               prefix_for_html,
2582                                               domain_id_to_secondary_features_maps );
2583                     }
2584                     else {
2585                         out.write( "<td></td>" );
2586                     }
2587                     if ( !go_id_to_term_map.containsKey( go_id ) ) {
2588                         throw new IllegalArgumentException( "GO-id [" + go_id + "] not found in GO-id to GO-term map" );
2589                     }
2590                     final GoTerm go_term = go_id_to_term_map.get( go_id );
2591                     if ( ( go_namespace_limit == null ) || go_namespace_limit.equals( go_term.getGoNameSpace() ) ) {
2592                         // final String top = GoUtils.getPenultimateGoTerm( go_term, go_id_to_term_map ).getName();
2593                         final String go_id_str = go_id.getId();
2594                         out.write( "<td>" );
2595                         out.write( "<a href=\"" + SurfacingConstants.AMIGO_LINK + go_id_str
2596                                 + "\" target=\"amigo_window\">" + go_id_str + "</a>" );
2597                         out.write( "</td><td>" );
2598                         out.write( go_term.getName() );
2599                         if ( domain_count == 2 ) {
2600                             out.write( " (" + d + ")" );
2601                         }
2602                         out.write( "</td><td>" );
2603                         // out.write( top );
2604                         // out.write( "</td><td>" );
2605                         out.write( "[" );
2606                         out.write( go_term.getGoNameSpace().toShortString() );
2607                         out.write( "]" );
2608                         out.write( "</td>" );
2609                         if ( all_go_ids != null ) {
2610                             all_go_ids.add( go_id );
2611                         }
2612                     }
2613                     else {
2614                         out.write( "<td>" );
2615                         out.write( "</td><td>" );
2616                         out.write( "</td><td>" );
2617                         out.write( "</td><td>" );
2618                         out.write( "</td>" );
2619                     }
2620                     out.write( "</tr>" );
2621                     out.write( SurfacingConstants.NL );
2622                 }
2623             }
2624         } //  for( int d = 0; d < domain_count; ++d )
2625         if ( !any_go_annotation_present ) {
2626             out.write( "<tr>" );
2627             writeDomainIdsToHtml( out, domain_0, domain_1, prefix_for_html, domain_id_to_secondary_features_maps );
2628             out.write( "<td>" );
2629             out.write( "</td><td>" );
2630             out.write( "</td><td>" );
2631             out.write( "</td><td>" );
2632             out.write( "</td>" );
2633             out.write( "</tr>" );
2634             out.write( SurfacingConstants.NL );
2635         }
2636     }
2637
2638     private static void writeDomainIdsToHtml( final Writer out,
2639                                               final String domain_0,
2640                                               final String domain_1,
2641                                               final String prefix_for_detailed_html,
2642                                               final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps )
2643             throws IOException {
2644         out.write( "<td>" );
2645         if ( !ForesterUtil.isEmpty( prefix_for_detailed_html ) ) {
2646             out.write( prefix_for_detailed_html );
2647             out.write( " " );
2648         }
2649         out.write( "<a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK + domain_0 + "\">" + domain_0 + "</a>" );
2650         out.write( "</td>" );
2651     }
2652
2653     private static void writeDomainsToIndividualFilePerTreeNode( final Writer individual_files_writer,
2654                                                                  final String domain_0,
2655                                                                  final String domain_1 ) throws IOException {
2656         individual_files_writer.write( domain_0 );
2657         individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2658         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
2659             individual_files_writer.write( domain_1 );
2660             individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2661         }
2662     }
2663
2664     private static void writePfamsToFile( final String outfile_name, final SortedSet<String> pfams ) {
2665         try {
2666             final Writer writer = new BufferedWriter( new FileWriter( new File( outfile_name ) ) );
2667             for( final String pfam : pfams ) {
2668                 writer.write( pfam );
2669                 writer.write( ForesterUtil.LINE_SEPARATOR );
2670             }
2671             writer.close();
2672             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote " + pfams.size() + " pfams to [" + outfile_name
2673                     + "]" );
2674         }
2675         catch ( final IOException e ) {
2676             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
2677         }
2678     }
2679
2680     private static void writeToNexus( final String outfile_name,
2681                                       final CharacterStateMatrix<BinaryStates> matrix,
2682                                       final Phylogeny phylogeny ) {
2683         if ( !( matrix instanceof BasicCharacterStateMatrix ) ) {
2684             throw new IllegalArgumentException( "can only write matrices of type [" + BasicCharacterStateMatrix.class
2685                     + "] to nexus" );
2686         }
2687         final BasicCharacterStateMatrix<BinaryStates> my_matrix = ( org.forester.evoinference.matrix.character.BasicCharacterStateMatrix<BinaryStates> ) matrix;
2688         final List<Phylogeny> phylogenies = new ArrayList<Phylogeny>( 1 );
2689         phylogenies.add( phylogeny );
2690         try {
2691             final BufferedWriter w = new BufferedWriter( new FileWriter( outfile_name ) );
2692             w.write( NexusConstants.NEXUS );
2693             w.write( ForesterUtil.LINE_SEPARATOR );
2694             my_matrix.writeNexusTaxaBlock( w );
2695             my_matrix.writeNexusBinaryChractersBlock( w );
2696             PhylogenyWriter.writeNexusTreesBlock( w, phylogenies, NH_CONVERSION_SUPPORT_VALUE_STYLE.NONE );
2697             w.flush();
2698             w.close();
2699             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote Nexus file: \"" + outfile_name + "\"" );
2700         }
2701         catch ( final IOException e ) {
2702             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2703         }
2704     }
2705
2706     private static void writeToNexus( final String outfile_name,
2707                                       final DomainParsimonyCalculator domain_parsimony,
2708                                       final Phylogeny phylogeny ) {
2709         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAINS,
2710                       domain_parsimony.createMatrixOfDomainPresenceOrAbsence(),
2711                       phylogeny );
2712         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAIN_COMBINATIONS,
2713                       domain_parsimony.createMatrixOfBinaryDomainCombinationPresenceOrAbsence(),
2714                       phylogeny );
2715     }
2716
2717     final static class DomainComparator implements Comparator<Domain> {
2718
2719         final private boolean _ascending;
2720
2721         public DomainComparator( final boolean ascending ) {
2722             _ascending = ascending;
2723         }
2724
2725         @Override
2726         public final int compare( final Domain d0, final Domain d1 ) {
2727             if ( d0.getFrom() < d1.getFrom() ) {
2728                 return _ascending ? -1 : 1;
2729             }
2730             else if ( d0.getFrom() > d1.getFrom() ) {
2731                 return _ascending ? 1 : -1;
2732             }
2733             return 0;
2734         }
2735     }
2736 }