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: www.phylosoft.org/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.List;
  43 import java.util.Map;
  44 import java.util.PriorityQueue;
  45 import java.util.Set;
  46 import java.util.SortedMap;
  47 import java.util.SortedSet;
  48 import java.util.TreeMap;
  49 import java.util.TreeSet;
  50 import java.util.regex.Matcher;
  51 import java.util.regex.Pattern;
  52
  53 import org.forester.application.surfacing;
  54 import org.forester.evoinference.distance.NeighborJoining;
  55 import org.forester.evoinference.matrix.character.BasicCharacterStateMatrix;
  56 import org.forester.evoinference.matrix.character.CharacterStateMatrix;
  57 import org.forester.evoinference.matrix.character.CharacterStateMatrix.BinaryStates;
  58 import org.forester.evoinference.matrix.character.CharacterStateMatrix.Format;
  59 import org.forester.evoinference.matrix.character.CharacterStateMatrix.GainLossStates;
  60 import org.forester.evoinference.matrix.distance.DistanceMatrix;
  61 import org.forester.go.GoId;
  62 import org.forester.go.GoNameSpace;
  63 import org.forester.go.GoTerm;
  64 import org.forester.go.PfamToGoMapping;
  65 import org.forester.io.parsers.nexus.NexusConstants;
  66 import org.forester.io.writers.PhylogenyWriter;
  67 import org.forester.phylogeny.Phylogeny;
  68 import org.forester.phylogeny.PhylogenyMethods;
  69 import org.forester.phylogeny.PhylogenyNode;
  70 import org.forester.phylogeny.data.BinaryCharacters;
  71 import org.forester.phylogeny.data.Confidence;
  72 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  73 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  74 import org.forester.surfacing.DomainSimilarityCalculator.GoAnnotationOutput;
  75 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  76 import org.forester.util.AsciiHistogram;
  77 import org.forester.util.BasicDescriptiveStatistics;
  78 import org.forester.util.BasicTable;
  79 import org.forester.util.BasicTableParser;
  80 import org.forester.util.DescriptiveStatistics;
  81 import org.forester.util.ForesterUtil;
  82
  83 public final class SurfacingUtil {
  84
  85     private final static NumberFormat       FORMATTER                        = new DecimalFormat( "0.0E0" );
  86     private final static NumberFormat       FORMATTER_3                      = new DecimalFormat( "0.000" );
  87     private static final Comparator<Domain> ASCENDING_CONFIDENCE_VALUE_ORDER = new Comparator<Domain>() {
  88
  89                                                                                  @Override
  90                                                                                  public int compare( final Domain d1,
  91                                                                                                      final Domain d2 ) {
  92                                                                                      if ( d1.getPerSequenceEvalue() < d2
  93                                                                                              .getPerSequenceEvalue() ) {
  94                                                                                          return -1;
  95                                                                                      }
  96                                                                                      else if ( d1
  97                                                                                              .getPerSequenceEvalue() > d2
  98                                                                                              .getPerSequenceEvalue() ) {
  99                                                                                          return 1;
 100                                                                                      }
 101                                                                                      else {
 102                                                                                          return d1.compareTo( d2 );
 103                                                                                      }
 104                                                                                  }
 105                                                                              };
 106     public final static Pattern             PATTERN_SP_STYLE_TAXONOMY        = Pattern.compile( "^[A-Z0-9]{3,5}$" );
 107
 108     private SurfacingUtil() {
 109         // Hidden constructor.
 110     }
 111
 112     public static void addAllBinaryDomainCombinationToSet( final GenomeWideCombinableDomains genome,
 113                                                            final SortedSet<BinaryDomainCombination> binary_domain_combinations ) {
 114         final SortedMap<DomainId, CombinableDomains> all_cd = genome.getAllCombinableDomainsIds();
 115         for( final DomainId domain_id : all_cd.keySet() ) {
 116             binary_domain_combinations.addAll( all_cd.get( domain_id ).toBinaryDomainCombinations() );
 117         }
 118     }
 119
 120     public static void addAllDomainIdsToSet( final GenomeWideCombinableDomains genome,
 121                                              final SortedSet<DomainId> domain_ids ) {
 122         final SortedSet<DomainId> domains = genome.getAllDomainIds();
 123         for( final DomainId domain : domains ) {
 124             domain_ids.add( domain );
 125         }
 126     }
 127
 128     public static void addHtmlHead( final Writer w, final String title ) throws IOException {
 129         w.write( SurfacingConstants.NL );
 130         w.write( "<head>" );
 131         w.write( "<title>" );
 132         w.write( title );
 133         w.write( "</title>" );
 134         w.write( SurfacingConstants.NL );
 135         w.write( "<style>" );
 136         w.write( SurfacingConstants.NL );
 137         w.write( "a:visited { color : #6633FF; text-decoration : none; }" );
 138         w.write( SurfacingConstants.NL );
 139         w.write( "a:link { color : #6633FF; text-decoration : none; }" );
 140         w.write( SurfacingConstants.NL );
 141         w.write( "a:active { color : #99FF00; text-decoration : none; }" );
 142         w.write( SurfacingConstants.NL );
 143         w.write( "a:hover { color : #FFFFFF; background-color : #99FF00; text-decoration : none; }" );
 144         w.write( SurfacingConstants.NL );
 145         w.write( "td { text-align: left; vertical-align: top; font-family: Verdana, Arial, Helvetica; font-size: 8pt}" );
 146         w.write( SurfacingConstants.NL );
 147         w.write( "h1 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 18pt; font-weight: bold }" );
 148         w.write( SurfacingConstants.NL );
 149         w.write( "h2 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 16pt; font-weight: bold }" );
 150         w.write( SurfacingConstants.NL );
 151         w.write( "</style>" );
 152         w.write( SurfacingConstants.NL );
 153         w.write( "</head>" );
 154         w.write( SurfacingConstants.NL );
 155     }
 156
 157     public static DescriptiveStatistics calculateDescriptiveStatisticsForMeanValues( final Set<DomainSimilarity> similarities ) {
 158         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 159         for( final DomainSimilarity similarity : similarities ) {
 160             stats.addValue( similarity.getMeanSimilarityScore() );
 161         }
 162         return stats;
 163     }
 164
 165     private static void calculateIndependentDomainCombinationGains( final Phylogeny local_phylogeny_l,
 166                                                                     final String outfilename_for_counts,
 167                                                                     final String outfilename_for_dc,
 168                                                                     final String outfilename_for_dc_for_go_mapping,
 169                                                                     final String outfilename_for_dc_for_go_mapping_unique ) {
 170         try {
 171             final BufferedWriter out_counts = new BufferedWriter( new FileWriter( outfilename_for_counts ) );
 172             final BufferedWriter out_dc = new BufferedWriter( new FileWriter( outfilename_for_dc ) );
 173             final BufferedWriter out_dc_for_go_mapping = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping ) );
 174             final BufferedWriter out_dc_for_go_mapping_unique = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping_unique ) );
 175             final SortedMap<String, Integer> dc_gain_counts = new TreeMap<String, Integer>();
 176             for( final PhylogenyNodeIterator it = local_phylogeny_l.iteratorPostorder(); it.hasNext(); ) {
 177                 final PhylogenyNode n = it.next();
 178                 final Set<String> gained_dc = n.getNodeData().getBinaryCharacters().getGainedCharacters();
 179                 for( final String dc : gained_dc ) {
 180                     if ( dc_gain_counts.containsKey( dc ) ) {
 181                         dc_gain_counts.put( dc, dc_gain_counts.get( dc ) + 1 );
 182                     }
 183                     else {
 184                         dc_gain_counts.put( dc, 1 );
 185                     }
 186                 }
 187             }
 188             final SortedMap<Integer, Integer> histogram = new TreeMap<Integer, Integer>();
 189             final SortedMap<Integer, StringBuilder> domain_lists = new TreeMap<Integer, StringBuilder>();
 190             final SortedMap<Integer, PriorityQueue<String>> domain_lists_go = new TreeMap<Integer, PriorityQueue<String>>();
 191             final SortedMap<Integer, SortedSet<String>> domain_lists_go_unique = new TreeMap<Integer, SortedSet<String>>();
 192             final Set<String> dcs = dc_gain_counts.keySet();
 193             for( final String dc : dcs ) {
 194                 final int count = dc_gain_counts.get( dc );
 195                 if ( histogram.containsKey( count ) ) {
 196                     histogram.put( count, histogram.get( count ) + 1 );
 197                     domain_lists.get( count ).append( ", " + dc );
 198                     domain_lists_go.get( count ).addAll( splitDomainCombination( dc ) );
 199                     domain_lists_go_unique.get( count ).addAll( splitDomainCombination( dc ) );
 200                 }
 201                 else {
 202                     histogram.put( count, 1 );
 203                     domain_lists.put( count, new StringBuilder( dc ) );
 204                     final PriorityQueue<String> q = new PriorityQueue<String>();
 205                     q.addAll( splitDomainCombination( dc ) );
 206                     domain_lists_go.put( count, q );
 207                     final SortedSet<String> set = new TreeSet<String>();
 208                     set.addAll( splitDomainCombination( dc ) );
 209                     domain_lists_go_unique.put( count, set );
 210                 }
 211             }
 212             final Set<Integer> histogram_keys = histogram.keySet();
 213             for( final Integer histogram_key : histogram_keys ) {
 214                 final int count = histogram.get( histogram_key );
 215                 final StringBuilder dc = domain_lists.get( histogram_key );
 216                 out_counts.write( histogram_key + "\t" + count + ForesterUtil.LINE_SEPARATOR );
 217                 out_dc.write( histogram_key + "\t" + dc + ForesterUtil.LINE_SEPARATOR );
 218                 out_dc_for_go_mapping.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
 219                 final Object[] sorted = domain_lists_go.get( histogram_key ).toArray();
 220                 Arrays.sort( sorted );
 221                 for( final Object domain : sorted ) {
 222                     out_dc_for_go_mapping.write( domain + ForesterUtil.LINE_SEPARATOR );
 223                 }
 224                 out_dc_for_go_mapping_unique.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
 225                 for( final String domain : domain_lists_go_unique.get( histogram_key ) ) {
 226                     out_dc_for_go_mapping_unique.write( domain + ForesterUtil.LINE_SEPARATOR );
 227                 }
 228             }
 229             out_counts.close();
 230             out_dc.close();
 231             out_dc_for_go_mapping.close();
 232             out_dc_for_go_mapping_unique.close();
 233         }
 234         catch ( final IOException e ) {
 235             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
 236         }
 237         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch counts to ["
 238                 + outfilename_for_counts + "]" );
 239         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch lists to ["
 240                 + outfilename_for_dc + "]" );
 241         ForesterUtil.programMessage( surfacing.PRG_NAME,
 242                                      "Wrote independent domain combination gains fitch lists to (for GO mapping) ["
 243                                              + outfilename_for_dc_for_go_mapping + "]" );
 244         ForesterUtil.programMessage( surfacing.PRG_NAME,
 245                                      "Wrote independent domain combination gains fitch lists to (for GO mapping, unique) ["
 246                                              + outfilename_for_dc_for_go_mapping_unique + "]" );
 247     }
 248
 249     public static int calculateOverlap( final Domain domain, final List<Boolean> covered_positions ) {
 250         int overlap_count = 0;
 251         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
 252             if ( ( i < covered_positions.size() ) && ( covered_positions.get( i ) == true ) ) {
 253                 ++overlap_count;
 254             }
 255         }
 256         return overlap_count;
 257     }
 258
 259     public static void checkForOutputFileWriteability( final File outfile ) {
 260         final String error = ForesterUtil.isWritableFile( outfile );
 261         if ( !ForesterUtil.isEmpty( error ) ) {
 262             ForesterUtil.fatalError( surfacing.PRG_NAME, error );
 263         }
 264     }
 265
 266     private static SortedSet<String> collectAllDomainsChangedOnSubtree( final PhylogenyNode subtree_root,
 267                                                                         final boolean get_gains ) {
 268         final SortedSet<String> domains = new TreeSet<String>();
 269         for( final PhylogenyNode descendant : PhylogenyMethods.getAllDescendants( subtree_root ) ) {
 270             final BinaryCharacters chars = descendant.getNodeData().getBinaryCharacters();
 271             if ( get_gains ) {
 272                 domains.addAll( chars.getGainedCharacters() );
 273             }
 274             else {
 275                 domains.addAll( chars.getLostCharacters() );
 276             }
 277         }
 278         return domains;
 279     }
 280
 281     public static void collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
 282                                                                                            final BinaryDomainCombination.DomainCombinationType dc_type,
 283                                                                                            final List<BinaryDomainCombination> all_binary_domains_combination_gained,
 284                                                                                            final boolean get_gains ) {
 285         final SortedSet<String> sorted_ids = new TreeSet<String>();
 286         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
 287             sorted_ids.add( matrix.getIdentifier( i ) );
 288         }
 289         for( final String id : sorted_ids ) {
 290             for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
 291                 if ( ( get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) )
 292                         || ( !get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.LOSS ) ) ) {
 293                     if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) {
 294                         all_binary_domains_combination_gained.add( AdjactantDirectedBinaryDomainCombination
 295                                 .createInstance( matrix.getCharacter( c ) ) );
 296                     }
 297                     else if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED ) {
 298                         all_binary_domains_combination_gained.add( DirectedBinaryDomainCombination
 299                                 .createInstance( matrix.getCharacter( c ) ) );
 300                     }
 301                     else {
 302                         all_binary_domains_combination_gained.add( BasicBinaryDomainCombination.createInstance( matrix
 303                                 .getCharacter( c ) ) );
 304                     }
 305                 }
 306             }
 307         }
 308     }
 309
 310     private static File createBaseDirForPerNodeDomainFiles( final String base_dir,
 311                                                             final boolean domain_combinations,
 312                                                             final CharacterStateMatrix.GainLossStates state,
 313                                                             final String outfile ) {
 314         File per_node_go_mapped_domain_gain_loss_files_base_dir = new File( new File( outfile ).getParent()
 315                 + ForesterUtil.FILE_SEPARATOR + base_dir );
 316         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 317             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 318         }
 319         if ( domain_combinations ) {
 320             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 321                     + ForesterUtil.FILE_SEPARATOR + "DC" );
 322         }
 323         else {
 324             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 325                     + ForesterUtil.FILE_SEPARATOR + "DOMAINS" );
 326         }
 327         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 328             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 329         }
 330         if ( state == GainLossStates.GAIN ) {
 331             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 332                     + ForesterUtil.FILE_SEPARATOR + "GAINS" );
 333         }
 334         else if ( state == GainLossStates.LOSS ) {
 335             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 336                     + ForesterUtil.FILE_SEPARATOR + "LOSSES" );
 337         }
 338         else {
 339             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 340                     + ForesterUtil.FILE_SEPARATOR + "PRESENT" );
 341         }
 342         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 343             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 344         }
 345         return per_node_go_mapped_domain_gain_loss_files_base_dir;
 346     }
 347
 348     public static Map<DomainId, List<GoId>> createDomainIdToGoIdMap( final List<PfamToGoMapping> pfam_to_go_mappings ) {
 349         final Map<DomainId, List<GoId>> domain_id_to_go_ids_map = new HashMap<DomainId, List<GoId>>( pfam_to_go_mappings
 350                 .size() );
 351         for( final PfamToGoMapping pfam_to_go : pfam_to_go_mappings ) {
 352             if ( !domain_id_to_go_ids_map.containsKey( pfam_to_go.getKey() ) ) {
 353                 domain_id_to_go_ids_map.put( pfam_to_go.getKey(), new ArrayList<GoId>() );
 354             }
 355             domain_id_to_go_ids_map.get( pfam_to_go.getKey() ).add( pfam_to_go.getValue() );
 356         }
 357         return domain_id_to_go_ids_map;
 358     }
 359
 360     public static Map<DomainId, Set<String>> createDomainIdToSecondaryFeaturesMap( final File secondary_features_map_file )
 361             throws IOException {
 362         final BasicTable<String> primary_table = BasicTableParser.parse( secondary_features_map_file, "\t" );
 363         final Map<DomainId, Set<String>> map = new TreeMap<DomainId, Set<String>>();
 364         for( int r = 0; r < primary_table.getNumberOfRows(); ++r ) {
 365             final DomainId domain_id = new DomainId( primary_table.getValue( 0, r ) );
 366             if ( !map.containsKey( domain_id ) ) {
 367                 map.put( domain_id, new HashSet<String>() );
 368             }
 369             map.get( domain_id ).add( primary_table.getValue( 1, r ) );
 370         }
 371         return map;
 372     }
 373
 374     public static Phylogeny createNjTreeBasedOnMatrixToFile( final File nj_tree_outfile, final DistanceMatrix distance ) {
 375         checkForOutputFileWriteability( nj_tree_outfile );
 376         final NeighborJoining nj = NeighborJoining.createInstance();
 377         final Phylogeny phylogeny = nj.execute( distance );
 378         phylogeny.setName( nj_tree_outfile.getName() );
 379         writePhylogenyToFile( phylogeny, nj_tree_outfile.toString() );
 380         return phylogeny;
 381     }
 382
 383     private static SortedSet<BinaryDomainCombination> createSetOfAllBinaryDomainCombinationsPerGenome( final GenomeWideCombinableDomains gwcd ) {
 384         final SortedMap<DomainId, CombinableDomains> cds = gwcd.getAllCombinableDomainsIds();
 385         final SortedSet<BinaryDomainCombination> binary_combinations = new TreeSet<BinaryDomainCombination>();
 386         for( final DomainId domain_id : cds.keySet() ) {
 387             final CombinableDomains cd = cds.get( domain_id );
 388             binary_combinations.addAll( cd.toBinaryDomainCombinations() );
 389         }
 390         return binary_combinations;
 391     }
 392
 393     public static void decoratePrintableDomainSimilarities( final SortedSet<DomainSimilarity> domain_similarities,
 394                                                             final Detailedness detailedness,
 395                                                             final GoAnnotationOutput go_annotation_output,
 396                                                             final Map<GoId, GoTerm> go_id_to_term_map,
 397                                                             final GoNameSpace go_namespace_limit ) {
 398         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) ) {
 399             throw new IllegalArgumentException( "attempt to use a GO namespace limit without a GO id to term map" );
 400         }
 401         for( final DomainSimilarity domain_similarity : domain_similarities ) {
 402             if ( domain_similarity instanceof PrintableDomainSimilarity ) {
 403                 final PrintableDomainSimilarity printable_domain_similarity = ( PrintableDomainSimilarity ) domain_similarity;
 404                 printable_domain_similarity.setDetailedness( detailedness );
 405                 printable_domain_similarity.setGoAnnotationOutput( go_annotation_output );
 406                 printable_domain_similarity.setGoIdToTermMap( go_id_to_term_map );
 407                 printable_domain_similarity.setGoNamespaceLimit( go_namespace_limit );
 408             }
 409         }
 410     }
 411
 412     public static void executeDomainLengthAnalysis( final String[][] input_file_properties,
 413                                                     final int number_of_genomes,
 414                                                     final DomainLengthsTable domain_lengths_table,
 415                                                     final File outfile ) throws IOException {
 416         final DecimalFormat df = new DecimalFormat( "#.00" );
 417         checkForOutputFileWriteability( outfile );
 418         final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
 419         out.write( "MEAN BASED STATISTICS PER SPECIES" );
 420         out.write( ForesterUtil.LINE_SEPARATOR );
 421         out.write( domain_lengths_table.createMeanBasedStatisticsPerSpeciesTable().toString() );
 422         out.write( ForesterUtil.LINE_SEPARATOR );
 423         out.write( ForesterUtil.LINE_SEPARATOR );
 424         final List<DomainLengths> domain_lengths_list = domain_lengths_table.getDomainLengthsList();
 425         out.write( "OUTLIER SPECIES PER DOMAIN (Z>=1.5)" );
 426         out.write( ForesterUtil.LINE_SEPARATOR );
 427         for( final DomainLengths domain_lengths : domain_lengths_list ) {
 428             final List<Species> species_list = domain_lengths.getMeanBasedOutlierSpecies( 1.5 );
 429             if ( species_list.size() > 0 ) {
 430                 out.write( domain_lengths.getDomainId() + "\t" );
 431                 for( final Species species : species_list ) {
 432                     out.write( species + "\t" );
 433                 }
 434                 out.write( ForesterUtil.LINE_SEPARATOR );
 435                 // DescriptiveStatistics stats_for_domain = domain_lengths
 436                 //         .calculateMeanBasedStatistics();
 437                 //AsciiHistogram histo = new AsciiHistogram( stats_for_domain );
 438                 //System.out.println( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 439             }
 440         }
 441         out.write( ForesterUtil.LINE_SEPARATOR );
 442         out.write( ForesterUtil.LINE_SEPARATOR );
 443         out.write( "OUTLIER SPECIES (Z 1.0)" );
 444         out.write( ForesterUtil.LINE_SEPARATOR );
 445         final DescriptiveStatistics stats_for_all_species = domain_lengths_table
 446                 .calculateMeanBasedStatisticsForAllSpecies();
 447         out.write( stats_for_all_species.asSummary() );
 448         out.write( ForesterUtil.LINE_SEPARATOR );
 449         final AsciiHistogram histo = new AsciiHistogram( stats_for_all_species );
 450         out.write( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 451         out.write( ForesterUtil.LINE_SEPARATOR );
 452         final double population_sd = stats_for_all_species.sampleStandardDeviation();
 453         final double population_mean = stats_for_all_species.arithmeticMean();
 454         for( final Species species : domain_lengths_table.getSpecies() ) {
 455             final double x = domain_lengths_table.calculateMeanBasedStatisticsForSpecies( species ).arithmeticMean();
 456             final double z = ( x - population_mean ) / population_sd;
 457             out.write( species + "\t" + z );
 458             out.write( ForesterUtil.LINE_SEPARATOR );
 459         }
 460         out.write( ForesterUtil.LINE_SEPARATOR );
 461         for( final Species species : domain_lengths_table.getSpecies() ) {
 462             final DescriptiveStatistics stats_for_species = domain_lengths_table
 463                     .calculateMeanBasedStatisticsForSpecies( species );
 464             final double x = stats_for_species.arithmeticMean();
 465             final double z = ( x - population_mean ) / population_sd;
 466             if ( ( z <= -1.0 ) || ( z >= 1.0 ) ) {
 467                 out.write( species + "\t" + df.format( z ) + "\t" + stats_for_species.asSummary() );
 468                 out.write( ForesterUtil.LINE_SEPARATOR );
 469             }
 470         }
 471         out.close();
 472         //        final List<HistogramData> histogram_datas = new ArrayList<HistogramData>();
 473         //        for( int i = 0; i < number_of_genomes; ++i ) {
 474         //            final Species species = new BasicSpecies( input_file_properties[ i ][ 0 ] );
 475         //            histogram_datas
 476         //                    .add( new HistogramData( species.toString(), domain_lengths_table
 477         //                            .calculateMeanBasedStatisticsForSpecies( species )
 478         //                            .getDataAsDoubleArray(), 5, 600, null, 60 ) );
 479         //        }
 480         //        final HistogramsFrame hf = new HistogramsFrame( histogram_datas );
 481         //        hf.setVisible( true );
 482         System.gc();
 483     }
 484
 485     /**
 486      *
 487      * @param all_binary_domains_combination_lost_fitch
 488      * @param consider_directedness_and_adjacency_for_bin_combinations
 489      * @param all_binary_domains_combination_gained if null ignored, otherwise this is to list all binary domain combinations
 490      * which were gained under unweighted (Fitch) parsimony.
 491      */
 492     public static void executeParsimonyAnalysis( final long random_number_seed_for_fitch_parsimony,
 493                                                  final boolean radomize_fitch_parsimony,
 494                                                  final String outfile_name,
 495                                                  final DomainParsimonyCalculator domain_parsimony,
 496                                                  final Phylogeny phylogeny,
 497                                                  final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 498                                                  final Map<GoId, GoTerm> go_id_to_term_map,
 499                                                  final GoNameSpace go_namespace_limit,
 500                                                  final String parameters_str,
 501                                                  final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
 502                                                  final SortedSet<DomainId> positive_filter,
 503                                                  final boolean output_binary_domain_combinations_for_graphs,
 504                                                  final List<BinaryDomainCombination> all_binary_domains_combination_gained_fitch,
 505                                                  final List<BinaryDomainCombination> all_binary_domains_combination_lost_fitch,
 506                                                  final BinaryDomainCombination.DomainCombinationType dc_type ) {
 507         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 508         final String date_time = ForesterUtil.getCurrentDateTime();
 509         final SortedSet<String> all_pfams_encountered = new TreeSet<String>();
 510         final SortedSet<String> all_pfams_gained_as_domains = new TreeSet<String>();
 511         final SortedSet<String> all_pfams_lost_as_domains = new TreeSet<String>();
 512         final SortedSet<String> all_pfams_gained_as_dom_combinations = new TreeSet<String>();
 513         final SortedSet<String> all_pfams_lost_as_dom_combinations = new TreeSet<String>();
 514         writeToNexus( outfile_name, domain_parsimony, phylogeny );
 515         // DOLLO DOMAINS
 516         // -------------
 517         Phylogeny local_phylogeny_l = phylogeny.copy();
 518         if ( ( positive_filter != null ) && ( positive_filter.size() > 0 ) ) {
 519             domain_parsimony.executeDolloParsimonyOnDomainPresence( positive_filter );
 520         }
 521         else {
 522             domain_parsimony.executeDolloParsimonyOnDomainPresence();
 523         }
 524         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 525                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 526         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 527                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 528         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 529                                                            CharacterStateMatrix.GainLossStates.GAIN,
 530                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_D,
 531                                                            sep,
 532                                                            ForesterUtil.LINE_SEPARATOR,
 533                                                            null );
 534         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 535                                                            CharacterStateMatrix.GainLossStates.LOSS,
 536                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_D,
 537                                                            sep,
 538                                                            ForesterUtil.LINE_SEPARATOR,
 539                                                            null );
 540         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 541                 + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_D, sep, ForesterUtil.LINE_SEPARATOR, null );
 542         //HTML:
 543         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 544                                        go_id_to_term_map,
 545                                        go_namespace_limit,
 546                                        false,
 547                                        domain_parsimony.getGainLossMatrix(),
 548                                        CharacterStateMatrix.GainLossStates.GAIN,
 549                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D,
 550                                        sep,
 551                                        ForesterUtil.LINE_SEPARATOR,
 552                                        "Dollo Parsimony | Gains | Domains",
 553                                        "+",
 554                                        domain_id_to_secondary_features_maps,
 555                                        all_pfams_encountered,
 556                                        all_pfams_gained_as_domains,
 557                                        "_dollo_gains_d" );
 558         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 559                                        go_id_to_term_map,
 560                                        go_namespace_limit,
 561                                        false,
 562                                        domain_parsimony.getGainLossMatrix(),
 563                                        CharacterStateMatrix.GainLossStates.LOSS,
 564                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D,
 565                                        sep,
 566                                        ForesterUtil.LINE_SEPARATOR,
 567                                        "Dollo Parsimony | Losses | Domains",
 568                                        "-",
 569                                        domain_id_to_secondary_features_maps,
 570                                        all_pfams_encountered,
 571                                        all_pfams_lost_as_domains,
 572                                        "_dollo_losses_d" );
 573         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 574                                        go_id_to_term_map,
 575                                        go_namespace_limit,
 576                                        false,
 577                                        domain_parsimony.getGainLossMatrix(),
 578                                        null,
 579                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D,
 580                                        sep,
 581                                        ForesterUtil.LINE_SEPARATOR,
 582                                        "Dollo Parsimony | Present | Domains",
 583                                        "",
 584                                        domain_id_to_secondary_features_maps,
 585                                        all_pfams_encountered,
 586                                        null,
 587                                        "_dollo_present_d" );
 588         preparePhylogeny( local_phylogeny_l,
 589                           domain_parsimony,
 590                           date_time,
 591                           "Dollo parsimony on domain presence/absence",
 592                           "dollo_on_domains_" + outfile_name,
 593                           parameters_str );
 594         SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 595                 + surfacing.DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 596         try {
 597             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, true, outfile_name, "_dollo_all_gains_d" );
 598             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, false, outfile_name, "_dollo_all_losses_d" );
 599         }
 600         catch ( final IOException e ) {
 601             e.printStackTrace();
 602             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
 603         }
 604         if ( domain_parsimony.calculateNumberOfBinaryDomainCombination() > 0 ) {
 605             // FITCH DOMAIN COMBINATIONS
 606             // -------------------------
 607             local_phylogeny_l = phylogeny.copy();
 608             String randomization = "no";
 609             if ( radomize_fitch_parsimony ) {
 610                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( random_number_seed_for_fitch_parsimony );
 611                 randomization = "yes, seed = " + random_number_seed_for_fitch_parsimony;
 612             }
 613             else {
 614                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( false );
 615             }
 616             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 617                     + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 618             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 619                     + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 620             SurfacingUtil
 621                     .writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 622                                                           CharacterStateMatrix.GainLossStates.GAIN,
 623                                                           outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_BC,
 624                                                           sep,
 625                                                           ForesterUtil.LINE_SEPARATOR,
 626                                                           null );
 627             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 628                                                                CharacterStateMatrix.GainLossStates.LOSS,
 629                                                                outfile_name
 630                                                                        + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_BC,
 631                                                                sep,
 632                                                                ForesterUtil.LINE_SEPARATOR,
 633                                                                null );
 634             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 635                     + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC, sep, ForesterUtil.LINE_SEPARATOR, null );
 636             if ( all_binary_domains_combination_gained_fitch != null ) {
 637                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 638                                                                                     dc_type,
 639                                                                                     all_binary_domains_combination_gained_fitch,
 640                                                                                     true );
 641             }
 642             if ( all_binary_domains_combination_lost_fitch != null ) {
 643                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 644                                                                                     dc_type,
 645                                                                                     all_binary_domains_combination_lost_fitch,
 646                                                                                     false );
 647             }
 648             if ( output_binary_domain_combinations_for_graphs ) {
 649                 SurfacingUtil
 650                         .writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( domain_parsimony
 651                                                                                                            .getGainLossMatrix(),
 652                                                                                                    null,
 653                                                                                                    outfile_name
 654                                                                                                            + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS,
 655                                                                                                    sep,
 656                                                                                                    ForesterUtil.LINE_SEPARATOR,
 657                                                                                                    BinaryDomainCombination.OutputFormat.DOT );
 658             }
 659             // HTML:
 660             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 661                                            go_id_to_term_map,
 662                                            go_namespace_limit,
 663                                            true,
 664                                            domain_parsimony.getGainLossMatrix(),
 665                                            CharacterStateMatrix.GainLossStates.GAIN,
 666                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC,
 667                                            sep,
 668                                            ForesterUtil.LINE_SEPARATOR,
 669                                            "Fitch Parsimony | Gains | Domain Combinations",
 670                                            "+",
 671                                            null,
 672                                            all_pfams_encountered,
 673                                            all_pfams_gained_as_dom_combinations,
 674                                            "_fitch_gains_dc" );
 675             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 676                                            go_id_to_term_map,
 677                                            go_namespace_limit,
 678                                            true,
 679                                            domain_parsimony.getGainLossMatrix(),
 680                                            CharacterStateMatrix.GainLossStates.LOSS,
 681                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC,
 682                                            sep,
 683                                            ForesterUtil.LINE_SEPARATOR,
 684                                            "Fitch Parsimony | Losses | Domain Combinations",
 685                                            "-",
 686                                            null,
 687                                            all_pfams_encountered,
 688                                            all_pfams_lost_as_dom_combinations,
 689                                            "_fitch_losses_dc" );
 690             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 691                                            go_id_to_term_map,
 692                                            go_namespace_limit,
 693                                            true,
 694                                            domain_parsimony.getGainLossMatrix(),
 695                                            null,
 696                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC,
 697                                            sep,
 698                                            ForesterUtil.LINE_SEPARATOR,
 699                                            "Fitch Parsimony | Present | Domain Combinations",
 700                                            "",
 701                                            null,
 702                                            all_pfams_encountered,
 703                                            null,
 704                                            "_fitch_present_dc" );
 705             writeAllEncounteredPfamsToFile( domain_id_to_go_ids_map,
 706                                             go_id_to_term_map,
 707                                             outfile_name,
 708                                             all_pfams_encountered );
 709             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX, all_pfams_gained_as_domains );
 710             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX, all_pfams_lost_as_domains );
 711             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DC_SUFFIX,
 712                               all_pfams_gained_as_dom_combinations );
 713             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DC_SUFFIX, all_pfams_lost_as_dom_combinations );
 714             preparePhylogeny( local_phylogeny_l,
 715                               domain_parsimony,
 716                               date_time,
 717                               "Fitch parsimony on binary domain combination presence/absence randomization: "
 718                                       + randomization,
 719                               "fitch_on_binary_domain_combinations_" + outfile_name,
 720                               parameters_str );
 721             SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 722                     + surfacing.BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH );
 723             calculateIndependentDomainCombinationGains( local_phylogeny_l, outfile_name
 724                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX, outfile_name
 725                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX, outfile_name
 726                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX, outfile_name
 727                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX );
 728         }
 729     }
 730
 731     public static void executeParsimonyAnalysisForSecondaryFeatures( final String outfile_name,
 732                                                                      final DomainParsimonyCalculator secondary_features_parsimony,
 733                                                                      final Phylogeny phylogeny,
 734                                                                      final String parameters_str,
 735                                                                      final Map<Species, MappingResults> mapping_results_map ) {
 736         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 737         final String date_time = ForesterUtil.getCurrentDateTime();
 738         System.out.println();
 739         writeToNexus( outfile_name + surfacing.NEXUS_SECONDARY_FEATURES,
 740                       secondary_features_parsimony.createMatrixOfSecondaryFeaturePresenceOrAbsence( null ),
 741                       phylogeny );
 742         final Phylogeny local_phylogeny_copy = phylogeny.copy();
 743         secondary_features_parsimony.executeDolloParsimonyOnSecondaryFeatures( mapping_results_map );
 744         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossMatrix(), outfile_name
 745                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 746         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossCountsMatrix(), outfile_name
 747                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 748         SurfacingUtil
 749                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 750                                                       CharacterStateMatrix.GainLossStates.GAIN,
 751                                                       outfile_name
 752                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES,
 753                                                       sep,
 754                                                       ForesterUtil.LINE_SEPARATOR,
 755                                                       null );
 756         SurfacingUtil
 757                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 758                                                       CharacterStateMatrix.GainLossStates.LOSS,
 759                                                       outfile_name
 760                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES,
 761                                                       sep,
 762                                                       ForesterUtil.LINE_SEPARATOR,
 763                                                       null );
 764         SurfacingUtil
 765                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 766                                                       null,
 767                                                       outfile_name
 768                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES,
 769                                                       sep,
 770                                                       ForesterUtil.LINE_SEPARATOR,
 771                                                       null );
 772         preparePhylogeny( local_phylogeny_copy,
 773                           secondary_features_parsimony,
 774                           date_time,
 775                           "Dollo parsimony on secondary feature presence/absence",
 776                           "dollo_on_secondary_features_" + outfile_name,
 777                           parameters_str );
 778         SurfacingUtil.writePhylogenyToFile( local_phylogeny_copy, outfile_name
 779                 + surfacing.SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 780     }
 781
 782     public static void extractProteinNames( final List<Protein> proteins,
 783                                             final List<DomainId> query_domain_ids_nc_order,
 784                                             final Writer out,
 785                                             final String separator ) throws IOException {
 786         for( final Protein protein : proteins ) {
 787             if ( protein.contains( query_domain_ids_nc_order, true ) ) {
 788                 out.write( protein.getSpecies().getSpeciesId() );
 789                 out.write( separator );
 790                 out.write( protein.getProteinId().getId() );
 791                 out.write( separator );
 792                 out.write( "[" );
 793                 final Set<DomainId> visited_domain_ids = new HashSet<DomainId>();
 794                 boolean first = true;
 795                 for( final Domain domain : protein.getProteinDomains() ) {
 796                     if ( !visited_domain_ids.contains( domain.getDomainId() ) ) {
 797                         visited_domain_ids.add( domain.getDomainId() );
 798                         if ( first ) {
 799                             first = false;
 800                         }
 801                         else {
 802                             out.write( " " );
 803                         }
 804                         out.write( domain.getDomainId().getId() );
 805                         out.write( " {" );
 806                         out.write( "" + domain.getTotalCount() );
 807                         out.write( "}" );
 808                     }
 809                 }
 810                 out.write( "]" );
 811                 out.write( separator );
 812                 if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 813                         .equals( SurfacingConstants.NONE ) ) ) {
 814                     out.write( protein.getDescription() );
 815                 }
 816                 out.write( separator );
 817                 if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 818                         .equals( SurfacingConstants.NONE ) ) ) {
 819                     out.write( protein.getAccession() );
 820                 }
 821                 out.write( SurfacingConstants.NL );
 822             }
 823         }
 824         out.flush();
 825     }
 826
 827     public static void extractProteinNames( final SortedMap<Species, List<Protein>> protein_lists_per_species,
 828                                             final DomainId domain_id,
 829                                             final Writer out,
 830                                             final String separator ) throws IOException {
 831         for( final Species species : protein_lists_per_species.keySet() ) {
 832             for( final Protein protein : protein_lists_per_species.get( species ) ) {
 833                 final List<Domain> domains = protein.getProteinDomains( domain_id );
 834                 if ( domains.size() > 0 ) {
 835                     final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 836                     for( final Domain domain : domains ) {
 837                         stats.addValue( domain.getPerSequenceEvalue() );
 838                     }
 839                     out.write( protein.getSpecies().getSpeciesId() );
 840                     out.write( separator );
 841                     out.write( protein.getProteinId().getId() );
 842                     out.write( separator );
 843                     out.write( "[" + FORMATTER.format( stats.median() ) + "]" );
 844                     out.write( separator );
 845                     if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 846                             .equals( SurfacingConstants.NONE ) ) ) {
 847                         out.write( protein.getDescription() );
 848                     }
 849                     out.write( separator );
 850                     if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 851                             .equals( SurfacingConstants.NONE ) ) ) {
 852                         out.write( protein.getAccession() );
 853                     }
 854                     out.write( SurfacingConstants.NL );
 855                 }
 856             }
 857         }
 858         out.flush();
 859     }
 860
 861     public static SortedSet<DomainId> getAllDomainIds( final List<GenomeWideCombinableDomains> gwcd_list ) {
 862         final SortedSet<DomainId> all_domains_ids = new TreeSet<DomainId>();
 863         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
 864             final Set<DomainId> all_domains = gwcd.getAllDomainIds();
 865             //    for( final Domain domain : all_domains ) {
 866             all_domains_ids.addAll( all_domains );
 867             //    }
 868         }
 869         return all_domains_ids;
 870     }
 871
 872     public static SortedMap<String, Integer> getDomainCounts( final List<Protein> protein_domain_collections ) {
 873         final SortedMap<String, Integer> map = new TreeMap<String, Integer>();
 874         for( final Protein protein_domain_collection : protein_domain_collections ) {
 875             for( final Object name : protein_domain_collection.getProteinDomains() ) {
 876                 final BasicDomain protein_domain = ( BasicDomain ) name;
 877                 final String id = protein_domain.getDomainId().getId();
 878                 if ( map.containsKey( id ) ) {
 879                     map.put( id, map.get( id ) + 1 );
 880                 }
 881                 else {
 882                     map.put( id, 1 );
 883                 }
 884             }
 885         }
 886         return map;
 887     }
 888
 889     public static int getNumberOfNodesLackingName( final Phylogeny p, final StringBuilder names ) {
 890         final PhylogenyNodeIterator it = p.iteratorPostorder();
 891         int c = 0;
 892         while ( it.hasNext() ) {
 893             final PhylogenyNode n = it.next();
 894             if ( ForesterUtil.isEmpty( n.getName() )
 895                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
 896                             .getScientificName() ) )
 897                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
 898                             .getCommonName() ) ) ) {
 899                 if ( n.getParent() != null ) {
 900                     names.append( " " );
 901                     names.append( n.getParent().getName() );
 902                 }
 903                 final List l = n.getAllExternalDescendants();
 904                 for( final Object object : l ) {
 905                     System.out.println( l.toString() );
 906                 }
 907                 ++c;
 908             }
 909         }
 910         return c;
 911     }
 912
 913     /**
 914      * Returns true is Domain domain falls in an uninterrupted stretch of
 915      * covered positions.
 916      *
 917      * @param domain
 918      * @param covered_positions
 919      * @return
 920      */
 921     public static boolean isEngulfed( final Domain domain, final List<Boolean> covered_positions ) {
 922         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
 923             if ( ( i >= covered_positions.size() ) || ( covered_positions.get( i ) != true ) ) {
 924                 return false;
 925             }
 926         }
 927         return true;
 928     }
 929
 930     public static void preparePhylogeny( final Phylogeny p,
 931                                          final DomainParsimonyCalculator domain_parsimony,
 932                                          final String date_time,
 933                                          final String method,
 934                                          final String name,
 935                                          final String parameters_str ) {
 936         domain_parsimony.decoratePhylogenyWithDomains( p );
 937         final StringBuilder desc = new StringBuilder();
 938         desc.append( "[Method: " + method + "] [Date: " + date_time + "] " );
 939         desc.append( "[Cost: " + domain_parsimony.getCost() + "] " );
 940         desc.append( "[Gains: " + domain_parsimony.getTotalGains() + "] " );
 941         desc.append( "[Losses: " + domain_parsimony.getTotalLosses() + "] " );
 942         desc.append( "[Unchanged: " + domain_parsimony.getTotalUnchanged() + "] " );
 943         desc.append( "[Parameters: " + parameters_str + "]" );
 944         p.setName( name );
 945         p.setDescription( desc.toString() );
 946         p.setConfidence( new Confidence( domain_parsimony.getCost(), "parsimony" ) );
 947         p.setRerootable( false );
 948         p.setRooted( true );
 949     }
 950
 951     /*
 952      * species | protein id | n-terminal domain | c-terminal domain | n-terminal domain per domain E-value | c-terminal domain per domain E-value
 953      *
 954      *
 955      */
 956     static public StringBuffer proteinToDomainCombinations( final Protein protein,
 957                                                             final String protein_id,
 958                                                             final String separator ) {
 959         final StringBuffer sb = new StringBuffer();
 960         if ( protein.getSpecies() == null ) {
 961             throw new IllegalArgumentException( "species must not be null" );
 962         }
 963         if ( ForesterUtil.isEmpty( protein.getSpecies().getSpeciesId() ) ) {
 964             throw new IllegalArgumentException( "species id must not be empty" );
 965         }
 966         final List<Domain> domains = protein.getProteinDomains();
 967         if ( domains.size() > 1 ) {
 968             final Map<String, Integer> counts = new HashMap<String, Integer>();
 969             for( final Domain domain : domains ) {
 970                 final String id = domain.getDomainId().getId();
 971                 if ( counts.containsKey( id ) ) {
 972                     counts.put( id, counts.get( id ) + 1 );
 973                 }
 974                 else {
 975                     counts.put( id, 1 );
 976                 }
 977             }
 978             final Set<String> dcs = new HashSet<String>();
 979             for( int i = 1; i < domains.size(); ++i ) {
 980                 for( int j = 0; j < i; ++j ) {
 981                     Domain domain_n = domains.get( i );
 982                     Domain domain_c = domains.get( j );
 983                     if ( domain_n.getFrom() > domain_c.getFrom() ) {
 984                         domain_n = domains.get( j );
 985                         domain_c = domains.get( i );
 986                     }
 987                     final String dc = domain_n.getDomainId().getId() + domain_c.getDomainId().getId();
 988                     if ( !dcs.contains( dc ) ) {
 989                         dcs.add( dc );
 990                         sb.append( protein.getSpecies() );
 991                         sb.append( separator );
 992                         sb.append( protein_id );
 993                         sb.append( separator );
 994                         sb.append( domain_n.getDomainId().getId() );
 995                         sb.append( separator );
 996                         sb.append( domain_c.getDomainId().getId() );
 997                         sb.append( separator );
 998                         sb.append( domain_n.getPerDomainEvalue() );
 999                         sb.append( separator );
1000                         sb.append( domain_c.getPerDomainEvalue() );
1001                         sb.append( separator );
1002                         sb.append( counts.get( domain_n.getDomainId().getId() ) );
1003                         sb.append( separator );
1004                         sb.append( counts.get( domain_c.getDomainId().getId() ) );
1005                         sb.append( ForesterUtil.LINE_SEPARATOR );
1006                     }
1007                 }
1008             }
1009         }
1010         else if ( domains.size() == 1 ) {
1011             sb.append( protein.getSpecies() );
1012             sb.append( separator );
1013             sb.append( protein_id );
1014             sb.append( separator );
1015             sb.append( domains.get( 0 ).getDomainId().getId() );
1016             sb.append( separator );
1017             sb.append( separator );
1018             sb.append( domains.get( 0 ).getPerDomainEvalue() );
1019             sb.append( separator );
1020             sb.append( separator );
1021             sb.append( 1 );
1022             sb.append( separator );
1023             sb.append( ForesterUtil.LINE_SEPARATOR );
1024         }
1025         else {
1026             sb.append( protein.getSpecies() );
1027             sb.append( separator );
1028             sb.append( protein_id );
1029             sb.append( separator );
1030             sb.append( separator );
1031             sb.append( separator );
1032             sb.append( separator );
1033             sb.append( separator );
1034             sb.append( separator );
1035             sb.append( ForesterUtil.LINE_SEPARATOR );
1036         }
1037         return sb;
1038     }
1039
1040     /**
1041      *
1042      * Example regarding engulfment: ------------0.1 ----------0.2 --0.3 =>
1043      * domain with 0.3 is ignored
1044      *
1045      * -----------0.1 ----------0.2 --0.3 => domain with 0.3 is ignored
1046      *
1047      *
1048      * ------------0.1 ----------0.3 --0.2 => domains with 0.3 and 0.2 are _not_
1049      * ignored
1050      *
1051      * @param max_allowed_overlap
1052      *            maximal allowed overlap (inclusive) to be still considered not
1053      *            overlapping (zero or negative value to allow any overlap)
1054      * @param remove_engulfed_domains
1055      *            to remove domains which are completely engulfed by coverage of
1056      *            domains with better support
1057      * @param protein
1058      * @return
1059      */
1060     public static Protein removeOverlappingDomains( final int max_allowed_overlap,
1061                                                     final boolean remove_engulfed_domains,
1062                                                     final Protein protein ) {
1063         final Protein pruned_protein = new BasicProtein( protein.getProteinId().getId(), protein.getSpecies()
1064                 .getSpeciesId() );
1065         final List<Domain> sorted = SurfacingUtil.sortDomainsWithAscendingConfidenceValues( protein );
1066         final List<Boolean> covered_positions = new ArrayList<Boolean>();
1067         for( final Domain domain : sorted ) {
1068             if ( ( ( max_allowed_overlap < 0 ) || ( SurfacingUtil.calculateOverlap( domain, covered_positions ) <= max_allowed_overlap ) )
1069                     && ( !remove_engulfed_domains || !isEngulfed( domain, covered_positions ) ) ) {
1070                 final int covered_positions_size = covered_positions.size();
1071                 for( int i = covered_positions_size; i < domain.getFrom(); ++i ) {
1072                     covered_positions.add( false );
1073                 }
1074                 final int new_covered_positions_size = covered_positions.size();
1075                 for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
1076                     if ( i < new_covered_positions_size ) {
1077                         covered_positions.set( i, true );
1078                     }
1079                     else {
1080                         covered_positions.add( true );
1081                     }
1082                 }
1083                 pruned_protein.addProteinDomain( domain );
1084             }
1085         }
1086         return pruned_protein;
1087     }
1088
1089     public static List<Domain> sortDomainsWithAscendingConfidenceValues( final Protein protein ) {
1090         final List<Domain> domains = new ArrayList<Domain>();
1091         for( final Domain d : protein.getProteinDomains() ) {
1092             domains.add( d );
1093         }
1094         Collections.sort( domains, SurfacingUtil.ASCENDING_CONFIDENCE_VALUE_ORDER );
1095         return domains;
1096     }
1097
1098     private static List<String> splitDomainCombination( final String dc ) {
1099         final String[] s = dc.split( "=" );
1100         if ( s.length != 2 ) {
1101             ForesterUtil.printErrorMessage( surfacing.PRG_NAME, "Stringyfied domain combination has illegal format: "
1102                     + dc );
1103             System.exit( -1 );
1104         }
1105         final List<String> l = new ArrayList<String>( 2 );
1106         l.add( s[ 0 ] );
1107         l.add( s[ 1 ] );
1108         return l;
1109     }
1110
1111     public static void writeAllDomainsChangedOnAllSubtrees( final Phylogeny p,
1112                                                             final boolean get_gains,
1113                                                             final String outdir,
1114                                                             final String suffix_for_filename ) throws IOException {
1115         CharacterStateMatrix.GainLossStates state = CharacterStateMatrix.GainLossStates.GAIN;
1116         if ( !get_gains ) {
1117             state = CharacterStateMatrix.GainLossStates.LOSS;
1118         }
1119         final File base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES,
1120                                                                   false,
1121                                                                   state,
1122                                                                   outdir );
1123         for( final PhylogenyNodeIterator it = p.iteratorPostorder(); it.hasNext(); ) {
1124             final PhylogenyNode node = it.next();
1125             if ( !node.isExternal() ) {
1126                 final SortedSet<String> domains = collectAllDomainsChangedOnSubtree( node, get_gains );
1127                 if ( domains.size() > 0 ) {
1128                     final Writer writer = ForesterUtil.createBufferedWriter( base_dir + ForesterUtil.FILE_SEPARATOR
1129                             + node.getName() + suffix_for_filename );
1130                     for( final String domain : domains ) {
1131                         writer.write( domain );
1132                         writer.write( ForesterUtil.LINE_SEPARATOR );
1133                     }
1134                     writer.close();
1135                 }
1136             }
1137         }
1138     }
1139
1140     private static void writeAllEncounteredPfamsToFile( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1141                                                         final Map<GoId, GoTerm> go_id_to_term_map,
1142                                                         final String outfile_name,
1143                                                         final SortedSet<String> all_pfams_encountered ) {
1144         final File all_pfams_encountered_file = new File( outfile_name + surfacing.ALL_PFAMS_ENCOUNTERED_SUFFIX );
1145         final File all_pfams_encountered_with_go_annotation_file = new File( outfile_name
1146                 + surfacing.ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX );
1147         final File encountered_pfams_summary_file = new File( outfile_name + surfacing.ENCOUNTERED_PFAMS_SUMMARY_SUFFIX );
1148         int biological_process_counter = 0;
1149         int cellular_component_counter = 0;
1150         int molecular_function_counter = 0;
1151         int pfams_with_mappings_counter = 0;
1152         int pfams_without_mappings_counter = 0;
1153         int pfams_without_mappings_to_bp_or_mf_counter = 0;
1154         int pfams_with_mappings_to_bp_or_mf_counter = 0;
1155         try {
1156             final Writer all_pfams_encountered_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_file ) );
1157             final Writer all_pfams_encountered_with_go_annotation_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_with_go_annotation_file ) );
1158             final Writer summary_writer = new BufferedWriter( new FileWriter( encountered_pfams_summary_file ) );
1159             summary_writer.write( "# Pfam to GO mapping summary" );
1160             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1161             summary_writer.write( "# Actual summary is at the end of this file." );
1162             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1163             summary_writer.write( "# Encountered Pfams without a GO mapping:" );
1164             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1165             for( final String pfam : all_pfams_encountered ) {
1166                 all_pfams_encountered_writer.write( pfam );
1167                 all_pfams_encountered_writer.write( ForesterUtil.LINE_SEPARATOR );
1168                 final DomainId domain_id = new DomainId( pfam );
1169                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1170                     ++pfams_with_mappings_counter;
1171                     all_pfams_encountered_with_go_annotation_writer.write( pfam );
1172                     all_pfams_encountered_with_go_annotation_writer.write( ForesterUtil.LINE_SEPARATOR );
1173                     final List<GoId> go_ids = domain_id_to_go_ids_map.get( domain_id );
1174                     boolean maps_to_bp = false;
1175                     boolean maps_to_cc = false;
1176                     boolean maps_to_mf = false;
1177                     for( final GoId go_id : go_ids ) {
1178                         final GoTerm go_term = go_id_to_term_map.get( go_id );
1179                         if ( go_term.getGoNameSpace().isBiologicalProcess() ) {
1180                             maps_to_bp = true;
1181                         }
1182                         else if ( go_term.getGoNameSpace().isCellularComponent() ) {
1183                             maps_to_cc = true;
1184                         }
1185                         else if ( go_term.getGoNameSpace().isMolecularFunction() ) {
1186                             maps_to_mf = true;
1187                         }
1188                     }
1189                     if ( maps_to_bp ) {
1190                         ++biological_process_counter;
1191                     }
1192                     if ( maps_to_cc ) {
1193                         ++cellular_component_counter;
1194                     }
1195                     if ( maps_to_mf ) {
1196                         ++molecular_function_counter;
1197                     }
1198                     if ( maps_to_bp || maps_to_mf ) {
1199                         ++pfams_with_mappings_to_bp_or_mf_counter;
1200                     }
1201                     else {
1202                         ++pfams_without_mappings_to_bp_or_mf_counter;
1203                     }
1204                 }
1205                 else {
1206                     ++pfams_without_mappings_to_bp_or_mf_counter;
1207                     ++pfams_without_mappings_counter;
1208                     summary_writer.write( pfam );
1209                     summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1210                 }
1211             }
1212             all_pfams_encountered_writer.close();
1213             all_pfams_encountered_with_go_annotation_writer.close();
1214             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + all_pfams_encountered.size()
1215                     + "] encountered Pfams to: \"" + all_pfams_encountered_file + "\"" );
1216             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + pfams_with_mappings_counter
1217                     + "] encountered Pfams with GO mappings to: \"" + all_pfams_encountered_with_go_annotation_file
1218                     + "\"" );
1219             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote summary (including all ["
1220                     + pfams_without_mappings_counter + "] encountered Pfams without GO mappings) to: \""
1221                     + encountered_pfams_summary_file + "\"" );
1222             ForesterUtil.programMessage( surfacing.PRG_NAME, "Sum of Pfams encountered                : "
1223                     + all_pfams_encountered.size() );
1224             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without a mapping                 : "
1225                     + pfams_without_mappings_counter + " ["
1226                     + ( 100 * pfams_without_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1227             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without mapping to proc. or func. : "
1228                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
1229                     + ( 100 * pfams_without_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1230             ForesterUtil.programMessage( surfacing.PRG_NAME,
1231                                          "Pfams with a mapping                    : " + pfams_with_mappings_counter
1232                                                  + " ["
1233                                                  + ( 100 * pfams_with_mappings_counter / all_pfams_encountered.size() )
1234                                                  + "%]" );
1235             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with a mapping to proc. or func.  : "
1236                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
1237                     + ( 100 * pfams_with_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1238             ForesterUtil.programMessage( surfacing.PRG_NAME,
1239                                          "Pfams with mapping to biological process: " + biological_process_counter
1240                                                  + " ["
1241                                                  + ( 100 * biological_process_counter / all_pfams_encountered.size() )
1242                                                  + "%]" );
1243             ForesterUtil.programMessage( surfacing.PRG_NAME,
1244                                          "Pfams with mapping to molecular function: " + molecular_function_counter
1245                                                  + " ["
1246                                                  + ( 100 * molecular_function_counter / all_pfams_encountered.size() )
1247                                                  + "%]" );
1248             ForesterUtil.programMessage( surfacing.PRG_NAME,
1249                                          "Pfams with mapping to cellular component: " + cellular_component_counter
1250                                                  + " ["
1251                                                  + ( 100 * cellular_component_counter / all_pfams_encountered.size() )
1252                                                  + "%]" );
1253             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1254             summary_writer.write( "# Sum of Pfams encountered                : " + all_pfams_encountered.size() );
1255             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1256             summary_writer.write( "# Pfams without a mapping                 : " + pfams_without_mappings_counter
1257                     + " [" + ( 100 * pfams_without_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1258             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1259             summary_writer.write( "# Pfams without mapping to proc. or func. : "
1260                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
1261                     + ( 100 * pfams_without_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1262             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1263             summary_writer.write( "# Pfams with a mapping                    : " + pfams_with_mappings_counter + " ["
1264                     + ( 100 * pfams_with_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1265             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1266             summary_writer.write( "# Pfams with a mapping to proc. or func.  : "
1267                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
1268                     + ( 100 * pfams_with_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1269             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1270             summary_writer.write( "# Pfams with mapping to biological process: " + biological_process_counter + " ["
1271                     + ( 100 * biological_process_counter / all_pfams_encountered.size() ) + "%]" );
1272             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1273             summary_writer.write( "# Pfams with mapping to molecular function: " + molecular_function_counter + " ["
1274                     + ( 100 * molecular_function_counter / all_pfams_encountered.size() ) + "%]" );
1275             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1276             summary_writer.write( "# Pfams with mapping to cellular component: " + cellular_component_counter + " ["
1277                     + ( 100 * cellular_component_counter / all_pfams_encountered.size() ) + "%]" );
1278             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1279             summary_writer.close();
1280         }
1281         catch ( final IOException e ) {
1282             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
1283         }
1284     }
1285
1286     public static void writeBinaryDomainCombinationsFileForGraphAnalysis( final String[][] input_file_properties,
1287                                                                           final File output_dir,
1288                                                                           final GenomeWideCombinableDomains gwcd,
1289                                                                           final int i,
1290                                                                           final GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1291         File dc_outfile_dot = new File( input_file_properties[ i ][ 0 ]
1292                 + surfacing.DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS );
1293         if ( output_dir != null ) {
1294             dc_outfile_dot = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile_dot );
1295         }
1296         checkForOutputFileWriteability( dc_outfile_dot );
1297         final SortedSet<BinaryDomainCombination> binary_combinations = createSetOfAllBinaryDomainCombinationsPerGenome( gwcd );
1298         try {
1299             final BufferedWriter out_dot = new BufferedWriter( new FileWriter( dc_outfile_dot ) );
1300             for( final BinaryDomainCombination bdc : binary_combinations ) {
1301                 out_dot.write( bdc.toGraphDescribingLanguage( BinaryDomainCombination.OutputFormat.DOT, null, null )
1302                         .toString() );
1303                 out_dot.write( SurfacingConstants.NL );
1304             }
1305             out_dot.close();
1306         }
1307         catch ( final IOException e ) {
1308             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1309         }
1310         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote binary domain combination for \""
1311                 + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1312                 + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile_dot + "\"" );
1313     }
1314
1315     public static void writeBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1316                                                             final CharacterStateMatrix.GainLossStates state,
1317                                                             final String filename,
1318                                                             final String indentifier_characters_separator,
1319                                                             final String character_separator,
1320                                                             final Map<String, String> descriptions ) {
1321         final File outfile = new File( filename );
1322         checkForOutputFileWriteability( outfile );
1323         final SortedSet<String> sorted_ids = new TreeSet<String>();
1324         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1325             sorted_ids.add( matrix.getIdentifier( i ) );
1326         }
1327         try {
1328             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1329             for( final String id : sorted_ids ) {
1330                 out.write( indentifier_characters_separator );
1331                 out.write( "#" + id );
1332                 out.write( indentifier_characters_separator );
1333                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1334                     // Not nice:
1335                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1336                     if ( ( matrix.getState( id, c ) == state )
1337                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1338                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1339                         out.write( matrix.getCharacter( c ) );
1340                         if ( ( descriptions != null ) && !descriptions.isEmpty()
1341                                 && descriptions.containsKey( matrix.getCharacter( c ) ) ) {
1342                             out.write( "\t" );
1343                             out.write( descriptions.get( matrix.getCharacter( c ) ) );
1344                         }
1345                         out.write( character_separator );
1346                     }
1347                 }
1348             }
1349             out.flush();
1350             out.close();
1351         }
1352         catch ( final IOException e ) {
1353             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1354         }
1355         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1356     }
1357
1358     public static void writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1359                                                                                                  final CharacterStateMatrix.GainLossStates state,
1360                                                                                                  final String filename,
1361                                                                                                  final String indentifier_characters_separator,
1362                                                                                                  final String character_separator,
1363                                                                                                  final BinaryDomainCombination.OutputFormat bc_output_format ) {
1364         final File outfile = new File( filename );
1365         checkForOutputFileWriteability( outfile );
1366         final SortedSet<String> sorted_ids = new TreeSet<String>();
1367         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1368             sorted_ids.add( matrix.getIdentifier( i ) );
1369         }
1370         try {
1371             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1372             for( final String id : sorted_ids ) {
1373                 out.write( indentifier_characters_separator );
1374                 out.write( "#" + id );
1375                 out.write( indentifier_characters_separator );
1376                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1377                     // Not nice:
1378                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1379                     if ( ( matrix.getState( id, c ) == state )
1380                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1381                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1382                         BinaryDomainCombination bdc = null;
1383                         try {
1384                             bdc = BasicBinaryDomainCombination.createInstance( matrix.getCharacter( c ) );
1385                         }
1386                         catch ( final Exception e ) {
1387                             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
1388                         }
1389                         out.write( bdc.toGraphDescribingLanguage( bc_output_format, null, null ).toString() );
1390                         out.write( character_separator );
1391                     }
1392                 }
1393             }
1394             out.flush();
1395             out.close();
1396         }
1397         catch ( final IOException e ) {
1398             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1399         }
1400         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1401     }
1402
1403     public static void writeBinaryStatesMatrixToList( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1404                                                       final Map<GoId, GoTerm> go_id_to_term_map,
1405                                                       final GoNameSpace go_namespace_limit,
1406                                                       final boolean domain_combinations,
1407                                                       final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1408                                                       final CharacterStateMatrix.GainLossStates state,
1409                                                       final String filename,
1410                                                       final String indentifier_characters_separator,
1411                                                       final String character_separator,
1412                                                       final String title_for_html,
1413                                                       final String prefix_for_html,
1414                                                       final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1415                                                       final SortedSet<String> all_pfams_encountered,
1416                                                       final SortedSet<String> pfams_gained_or_lost,
1417                                                       final String suffix_for_per_node_events_file ) {
1418         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1419             throw new IllegalArgumentException( "attempt to use GO namespace limit without a GO-id to term map" );
1420         }
1421         else if ( ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) ) {
1422             throw new IllegalArgumentException( "attempt to output detailed HTML without a Pfam to GO map" );
1423         }
1424         else if ( ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1425             throw new IllegalArgumentException( "attempt to output detailed HTML without a GO-id to term map" );
1426         }
1427         final File outfile = new File( filename );
1428         checkForOutputFileWriteability( outfile );
1429         final SortedSet<String> sorted_ids = new TreeSet<String>();
1430         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1431             sorted_ids.add( matrix.getIdentifier( i ) );
1432         }
1433         try {
1434             final Writer out = new BufferedWriter( new FileWriter( outfile ) );
1435             final File per_node_go_mapped_domain_gain_loss_files_base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES,
1436                                                                                                                 domain_combinations,
1437                                                                                                                 state,
1438                                                                                                                 filename );
1439             Writer per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1440             File per_node_go_mapped_domain_gain_loss_outfile = null;
1441             int per_node_counter = 0;
1442             out.write( "<html>" );
1443             out.write( SurfacingConstants.NL );
1444             addHtmlHead( out, title_for_html );
1445             out.write( SurfacingConstants.NL );
1446             out.write( "<body>" );
1447             out.write( SurfacingConstants.NL );
1448             out.write( "<h1>" );
1449             out.write( SurfacingConstants.NL );
1450             out.write( title_for_html );
1451             out.write( SurfacingConstants.NL );
1452             out.write( "</h1>" );
1453             out.write( SurfacingConstants.NL );
1454             out.write( "<table>" );
1455             out.write( SurfacingConstants.NL );
1456             for( final String id : sorted_ids ) {
1457                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1458                 if ( matcher.matches() ) {
1459                     continue;
1460                 }
1461                 out.write( "<tr>" );
1462                 out.write( "<td>" );
1463                 out.write( "<a href=\"#" + id + "\">" + id + "</a>" );
1464                 out.write( "</td>" );
1465                 out.write( "</tr>" );
1466                 out.write( SurfacingConstants.NL );
1467             }
1468             out.write( "</table>" );
1469             out.write( SurfacingConstants.NL );
1470             for( final String id : sorted_ids ) {
1471                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1472                 if ( matcher.matches() ) {
1473                     continue;
1474                 }
1475                 out.write( SurfacingConstants.NL );
1476                 out.write( "<h2>" );
1477                 out.write( "<a name=\"" + id + "\">" + id + "</a>" );
1478                 writeTaxonomyLinks( out, id );
1479                 out.write( "</h2>" );
1480                 out.write( SurfacingConstants.NL );
1481                 out.write( "<table>" );
1482                 out.write( SurfacingConstants.NL );
1483                 out.write( "<tr>" );
1484                 out.write( "<td><b>" );
1485                 out.write( "Pfam domain(s)" );
1486                 out.write( "</b></td><td><b>" );
1487                 out.write( "GO term acc" );
1488                 out.write( "</b></td><td><b>" );
1489                 out.write( "GO term" );
1490                 out.write( "</b></td><td><b>" );
1491                 out.write( "GO namespace" );
1492                 out.write( "</b></td>" );
1493                 out.write( "</tr>" );
1494                 out.write( SurfacingConstants.NL );
1495                 out.write( "</tr>" );
1496                 out.write( SurfacingConstants.NL );
1497                 per_node_counter = 0;
1498                 if ( matrix.getNumberOfCharacters() > 0 ) {
1499                     per_node_go_mapped_domain_gain_loss_outfile = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
1500                             + ForesterUtil.FILE_SEPARATOR + id + suffix_for_per_node_events_file );
1501                     SurfacingUtil.checkForOutputFileWriteability( per_node_go_mapped_domain_gain_loss_outfile );
1502                     per_node_go_mapped_domain_gain_loss_outfile_writer = ForesterUtil
1503                             .createBufferedWriter( per_node_go_mapped_domain_gain_loss_outfile );
1504                 }
1505                 else {
1506                     per_node_go_mapped_domain_gain_loss_outfile = null;
1507                     per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1508                 }
1509                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1510                     // Not nice:
1511                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1512                     if ( ( matrix.getState( id, c ) == state )
1513                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) || ( matrix
1514                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) ) ) ) {
1515                         final String character = matrix.getCharacter( c );
1516                         String domain_0 = "";
1517                         String domain_1 = "";
1518                         if ( character.indexOf( BinaryDomainCombination.SEPARATOR ) > 0 ) {
1519                             final String[] s = character.split( BinaryDomainCombination.SEPARATOR );
1520                             if ( s.length != 2 ) {
1521                                 throw new AssertionError( "this should not have happened: unexpected format for domain combination: ["
1522                                         + character + "]" );
1523                             }
1524                             domain_0 = s[ 0 ];
1525                             domain_1 = s[ 1 ];
1526                         }
1527                         else {
1528                             domain_0 = character;
1529                         }
1530                         writeDomainData( domain_id_to_go_ids_map,
1531                                          go_id_to_term_map,
1532                                          go_namespace_limit,
1533                                          out,
1534                                          domain_0,
1535                                          domain_1,
1536                                          prefix_for_html,
1537                                          character_separator,
1538                                          domain_id_to_secondary_features_maps,
1539                                          null );
1540                         all_pfams_encountered.add( domain_0 );
1541                         if ( pfams_gained_or_lost != null ) {
1542                             pfams_gained_or_lost.add( domain_0 );
1543                         }
1544                         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
1545                             all_pfams_encountered.add( domain_1 );
1546                             if ( pfams_gained_or_lost != null ) {
1547                                 pfams_gained_or_lost.add( domain_1 );
1548                             }
1549                         }
1550                         if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1551                             writeDomainsToIndividualFilePerTreeNode( per_node_go_mapped_domain_gain_loss_outfile_writer,
1552                                                                      domain_0,
1553                                                                      domain_1 );
1554                             per_node_counter++;
1555                         }
1556                     }
1557                 }
1558                 if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1559                     per_node_go_mapped_domain_gain_loss_outfile_writer.close();
1560                     if ( per_node_counter < 1 ) {
1561                         per_node_go_mapped_domain_gain_loss_outfile.delete();
1562                     }
1563                     per_node_counter = 0;
1564                 }
1565                 out.write( "</table>" );
1566                 out.write( SurfacingConstants.NL );
1567                 out.write( "<hr>" );
1568                 out.write( SurfacingConstants.NL );
1569             } // for( final String id : sorted_ids ) {
1570             out.write( "</body>" );
1571             out.write( SurfacingConstants.NL );
1572             out.write( "</html>" );
1573             out.write( SurfacingConstants.NL );
1574             out.flush();
1575             out.close();
1576         }
1577         catch ( final IOException e ) {
1578             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1579         }
1580         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters detailed HTML list: \"" + filename + "\"" );
1581     }
1582
1583     public static void writeBinaryStatesMatrixToListORIGIG( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1584                                                             final Map<GoId, GoTerm> go_id_to_term_map,
1585                                                             final GoNameSpace go_namespace_limit,
1586                                                             final boolean domain_combinations,
1587                                                             final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1588                                                             final CharacterStateMatrix.GainLossStates state,
1589                                                             final String filename,
1590                                                             final String indentifier_characters_separator,
1591                                                             final String character_separator,
1592                                                             final String title_for_html,
1593                                                             final String prefix_for_html,
1594                                                             final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1595                                                             final SortedSet<String> all_pfams_encountered,
1596                                                             final SortedSet<String> pfams_gained_or_lost,
1597                                                             final String suffix_for_per_node_events_file ) {
1598         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1599             throw new IllegalArgumentException( "attempt to use GO namespace limit without a GO-id to term map" );
1600         }
1601         else if ( ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) ) {
1602             throw new IllegalArgumentException( "attempt to output detailed HTML without a Pfam to GO map" );
1603         }
1604         else if ( ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1605             throw new IllegalArgumentException( "attempt to output detailed HTML without a GO-id to term map" );
1606         }
1607         final File outfile = new File( filename );
1608         checkForOutputFileWriteability( outfile );
1609         final SortedSet<String> sorted_ids = new TreeSet<String>();
1610         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1611             sorted_ids.add( matrix.getIdentifier( i ) );
1612         }
1613         try {
1614             final Writer out = new BufferedWriter( new FileWriter( outfile ) );
1615             final File per_node_go_mapped_domain_gain_loss_files_base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES,
1616                                                                                                                 domain_combinations,
1617                                                                                                                 state,
1618                                                                                                                 filename );
1619             Writer per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1620             File per_node_go_mapped_domain_gain_loss_outfile = null;
1621             int per_node_counter = 0;
1622             out.write( "<html>" );
1623             out.write( SurfacingConstants.NL );
1624             addHtmlHead( out, title_for_html );
1625             out.write( SurfacingConstants.NL );
1626             out.write( "<body>" );
1627             out.write( SurfacingConstants.NL );
1628             out.write( "<h1>" );
1629             out.write( SurfacingConstants.NL );
1630             out.write( title_for_html );
1631             out.write( SurfacingConstants.NL );
1632             out.write( "</h1>" );
1633             out.write( SurfacingConstants.NL );
1634             out.write( "<table>" );
1635             out.write( SurfacingConstants.NL );
1636             for( final String id : sorted_ids ) {
1637                 out.write( "<tr>" );
1638                 out.write( "<td>" );
1639                 out.write( "<a href=\"#" + id + "\">" + id + "</a>" );
1640                 writeTaxonomyLinks( out, id );
1641                 out.write( "</td>" );
1642                 out.write( "</tr>" );
1643                 out.write( SurfacingConstants.NL );
1644             }
1645             out.write( "</table>" );
1646             out.write( SurfacingConstants.NL );
1647             for( final String id : sorted_ids ) {
1648                 out.write( SurfacingConstants.NL );
1649                 out.write( "<h2>" );
1650                 out.write( "<a name=\"" + id + "\">" + id + "</a>" );
1651                 writeTaxonomyLinks( out, id );
1652                 out.write( "</h2>" );
1653                 out.write( SurfacingConstants.NL );
1654                 out.write( "<table>" );
1655                 out.write( SurfacingConstants.NL );
1656                 out.write( "<tr>" );
1657                 out.write( "<td><b>" );
1658                 out.write( "Pfam domain(s)" );
1659                 out.write( "</b></td><td><b>" );
1660                 out.write( "GO term acc" );
1661                 out.write( "</b></td><td><b>" );
1662                 out.write( "GO term" );
1663                 out.write( "</b></td><td><b>" );
1664                 out.write( "Penultimate GO term" );
1665                 out.write( "</b></td><td><b>" );
1666                 out.write( "GO namespace" );
1667                 out.write( "</b></td>" );
1668                 out.write( "</tr>" );
1669                 out.write( SurfacingConstants.NL );
1670                 out.write( "</tr>" );
1671                 out.write( SurfacingConstants.NL );
1672                 per_node_counter = 0;
1673                 if ( matrix.getNumberOfCharacters() > 0 ) {
1674                     per_node_go_mapped_domain_gain_loss_outfile = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
1675                             + ForesterUtil.FILE_SEPARATOR + id + suffix_for_per_node_events_file );
1676                     SurfacingUtil.checkForOutputFileWriteability( per_node_go_mapped_domain_gain_loss_outfile );
1677                     per_node_go_mapped_domain_gain_loss_outfile_writer = ForesterUtil
1678                             .createBufferedWriter( per_node_go_mapped_domain_gain_loss_outfile );
1679                 }
1680                 else {
1681                     per_node_go_mapped_domain_gain_loss_outfile = null;
1682                     per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1683                 }
1684                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1685                     // Not nice:
1686                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1687                     if ( ( matrix.getState( id, c ) == state )
1688                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) || ( matrix
1689                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) ) ) ) {
1690                         final String character = matrix.getCharacter( c );
1691                         String domain_0 = "";
1692                         String domain_1 = "";
1693                         if ( character.indexOf( BinaryDomainCombination.SEPARATOR ) > 0 ) {
1694                             final String[] s = character.split( BinaryDomainCombination.SEPARATOR );
1695                             if ( s.length != 2 ) {
1696                                 throw new AssertionError( "this should not have happened: unexpected format for domain combination: ["
1697                                         + character + "]" );
1698                             }
1699                             domain_0 = s[ 0 ];
1700                             domain_1 = s[ 1 ];
1701                         }
1702                         else {
1703                             domain_0 = character;
1704                         }
1705                         writeDomainData( domain_id_to_go_ids_map,
1706                                          go_id_to_term_map,
1707                                          go_namespace_limit,
1708                                          out,
1709                                          domain_0,
1710                                          domain_1,
1711                                          prefix_for_html,
1712                                          character_separator,
1713                                          domain_id_to_secondary_features_maps,
1714                                          null );
1715                         all_pfams_encountered.add( domain_0 );
1716                         if ( pfams_gained_or_lost != null ) {
1717                             pfams_gained_or_lost.add( domain_0 );
1718                         }
1719                         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
1720                             all_pfams_encountered.add( domain_1 );
1721                             if ( pfams_gained_or_lost != null ) {
1722                                 pfams_gained_or_lost.add( domain_1 );
1723                             }
1724                         }
1725                         if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1726                             writeDomainsToIndividualFilePerTreeNode( per_node_go_mapped_domain_gain_loss_outfile_writer,
1727                                                                      domain_0,
1728                                                                      domain_1 );
1729                             per_node_counter++;
1730                         }
1731                     }
1732                 }
1733                 if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1734                     per_node_go_mapped_domain_gain_loss_outfile_writer.close();
1735                     if ( per_node_counter < 1 ) {
1736                         per_node_go_mapped_domain_gain_loss_outfile.delete();
1737                     }
1738                     per_node_counter = 0;
1739                 }
1740                 out.write( "</table>" );
1741                 out.write( SurfacingConstants.NL );
1742                 out.write( "<hr>" );
1743                 out.write( SurfacingConstants.NL );
1744             } // for( final String id : sorted_ids ) {
1745             out.write( "</body>" );
1746             out.write( SurfacingConstants.NL );
1747             out.write( "</html>" );
1748             out.write( SurfacingConstants.NL );
1749             out.flush();
1750             out.close();
1751         }
1752         catch ( final IOException e ) {
1753             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1754         }
1755         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters detailed HTML list: \"" + filename + "\"" );
1756     }
1757
1758     public static void writeDomainCombinationsCountsFile( final String[][] input_file_properties,
1759                                                           final File output_dir,
1760                                                           final Writer per_genome_domain_promiscuity_statistics_writer,
1761                                                           final GenomeWideCombinableDomains gwcd,
1762                                                           final int i,
1763                                                           final GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1764         File dc_outfile = new File( input_file_properties[ i ][ 0 ]
1765                 + surfacing.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
1766         if ( output_dir != null ) {
1767             dc_outfile = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile );
1768         }
1769         checkForOutputFileWriteability( dc_outfile );
1770         try {
1771             final BufferedWriter out = new BufferedWriter( new FileWriter( dc_outfile ) );
1772             out.write( gwcd.toStringBuilder( dc_sort_order ).toString() );
1773             out.close();
1774         }
1775         catch ( final IOException e ) {
1776             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1777         }
1778         final DescriptiveStatistics stats = gwcd.getPerGenomeDomainPromiscuityStatistics();
1779         try {
1780             per_genome_domain_promiscuity_statistics_writer.write( input_file_properties[ i ][ 0 ] + "\t" );
1781             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.arithmeticMean() ) + "\t" );
1782             if ( stats.getN() < 2 ) {
1783                 per_genome_domain_promiscuity_statistics_writer.write( "n/a" + "\t" );
1784             }
1785             else {
1786                 per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats
1787                         .sampleStandardDeviation() ) + "\t" );
1788             }
1789             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.median() ) + "\t" );
1790             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMin() + "\t" );
1791             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMax() + "\t" );
1792             per_genome_domain_promiscuity_statistics_writer.write( stats.getN() + "\t" );
1793             final SortedSet<DomainId> mpds = gwcd.getMostPromiscuosDomain();
1794             for( final DomainId mpd : mpds ) {
1795                 per_genome_domain_promiscuity_statistics_writer.write( mpd.getId() + " " );
1796             }
1797             per_genome_domain_promiscuity_statistics_writer.write( ForesterUtil.LINE_SEPARATOR );
1798         }
1799         catch ( final IOException e ) {
1800             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1801         }
1802         if ( input_file_properties[ i ].length == 3 ) {
1803             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1804                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1805                     + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile + "\"" );
1806         }
1807         else {
1808             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1809                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ") to: \""
1810                     + dc_outfile + "\"" );
1811         }
1812     }
1813
1814     private static void writeDomainData( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1815                                          final Map<GoId, GoTerm> go_id_to_term_map,
1816                                          final GoNameSpace go_namespace_limit,
1817                                          final Writer out,
1818                                          final String domain_0,
1819                                          final String domain_1,
1820                                          final String prefix_for_html,
1821                                          final String character_separator_for_non_html_output,
1822                                          final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1823                                          final Set<GoId> all_go_ids ) throws IOException {
1824         boolean any_go_annotation_present = false;
1825         boolean first_has_no_go = false;
1826         int domain_count = 2; // To distinguish between domains and binary domain combinations.
1827         if ( ForesterUtil.isEmpty( domain_1 ) ) {
1828             domain_count = 1;
1829         }
1830         // The following has a difficult to understand logic.
1831         for( int d = 0; d < domain_count; ++d ) {
1832             List<GoId> go_ids = null;
1833             boolean go_annotation_present = false;
1834             if ( d == 0 ) {
1835                 final DomainId domain_id = new DomainId( domain_0 );
1836                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1837                     go_annotation_present = true;
1838                     any_go_annotation_present = true;
1839                     go_ids = domain_id_to_go_ids_map.get( domain_id );
1840                 }
1841                 else {
1842                     first_has_no_go = true;
1843                 }
1844             }
1845             else {
1846                 final DomainId domain_id = new DomainId( domain_1 );
1847                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1848                     go_annotation_present = true;
1849                     any_go_annotation_present = true;
1850                     go_ids = domain_id_to_go_ids_map.get( domain_id );
1851                 }
1852             }
1853             if ( go_annotation_present ) {
1854                 boolean first = ( ( d == 0 ) || ( ( d == 1 ) && first_has_no_go ) );
1855                 for( final GoId go_id : go_ids ) {
1856                     out.write( "<tr>" );
1857                     if ( first ) {
1858                         first = false;
1859                         writeDomainIdsToHtml( out,
1860                                               domain_0,
1861                                               domain_1,
1862                                               prefix_for_html,
1863                                               domain_id_to_secondary_features_maps );
1864                     }
1865                     else {
1866                         out.write( "<td></td>" );
1867                     }
1868                     if ( !go_id_to_term_map.containsKey( go_id ) ) {
1869                         throw new IllegalArgumentException( "GO-id [" + go_id + "] not found in GO-id to GO-term map" );
1870                     }
1871                     final GoTerm go_term = go_id_to_term_map.get( go_id );
1872                     if ( ( go_namespace_limit == null ) || go_namespace_limit.equals( go_term.getGoNameSpace() ) ) {
1873                         // final String top = GoUtils.getPenultimateGoTerm( go_term, go_id_to_term_map ).getName();
1874                         final String go_id_str = go_id.getId();
1875                         out.write( "<td>" );
1876                         out.write( "<a href=\"" + SurfacingConstants.AMIGO_LINK + go_id_str
1877                                 + "\" target=\"amigo_window\">" + go_id_str + "</a>" );
1878                         out.write( "</td><td>" );
1879                         out.write( go_term.getName() );
1880                         if ( domain_count == 2 ) {
1881                             out.write( " (" + d + ")" );
1882                         }
1883                         out.write( "</td><td>" );
1884                         // out.write( top );
1885                         // out.write( "</td><td>" );
1886                         out.write( "[" );
1887                         out.write( go_term.getGoNameSpace().toShortString() );
1888                         out.write( "]" );
1889                         out.write( "</td>" );
1890                         if ( all_go_ids != null ) {
1891                             all_go_ids.add( go_id );
1892                         }
1893                     }
1894                     else {
1895                         out.write( "<td>" );
1896                         out.write( "</td><td>" );
1897                         out.write( "</td><td>" );
1898                         out.write( "</td><td>" );
1899                         out.write( "</td>" );
1900                     }
1901                     out.write( "</tr>" );
1902                     out.write( SurfacingConstants.NL );
1903                 }
1904             }
1905         } //  for( int d = 0; d < domain_count; ++d )
1906         if ( !any_go_annotation_present ) {
1907             out.write( "<tr>" );
1908             writeDomainIdsToHtml( out, domain_0, domain_1, prefix_for_html, domain_id_to_secondary_features_maps );
1909             out.write( "<td>" );
1910             out.write( "</td><td>" );
1911             out.write( "</td><td>" );
1912             out.write( "</td><td>" );
1913             out.write( "</td>" );
1914             out.write( "</tr>" );
1915             out.write( SurfacingConstants.NL );
1916         }
1917     }
1918
1919     private static void writeDomainIdsToHtml( final Writer out,
1920                                               final String domain_0,
1921                                               final String domain_1,
1922                                               final String prefix_for_detailed_html,
1923                                               final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps )
1924             throws IOException {
1925         out.write( "<td>" );
1926         if ( !ForesterUtil.isEmpty( prefix_for_detailed_html ) ) {
1927             out.write( prefix_for_detailed_html );
1928             out.write( " " );
1929         }
1930         out.write( "<a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK + domain_0 + "\">" + domain_0 + "</a>" );
1931         out.write( "</td>" );
1932     }
1933
1934     public static DescriptiveStatistics writeDomainSimilaritiesToFile( final StringBuilder html_desc,
1935                                                                        final StringBuilder html_title,
1936                                                                        final Writer w,
1937                                                                        final SortedSet<DomainSimilarity> similarities,
1938                                                                        final boolean treat_as_binary,
1939                                                                        final List<Species> species_order,
1940                                                                        final PrintableDomainSimilarity.PRINT_OPTION print_option,
1941                                                                        final DomainSimilarity.DomainSimilaritySortField sort_field,
1942                                                                        final DomainSimilarity.DomainSimilarityScoring scoring,
1943                                                                        final boolean verbose ) throws IOException {
1944         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
1945         String histogram_title = null;
1946         switch ( sort_field ) {
1947             case ABS_MAX_COUNTS_DIFFERENCE:
1948                 if ( treat_as_binary ) {
1949                     histogram_title = "absolute counts difference:";
1950                 }
1951                 else {
1952                     histogram_title = "absolute (maximal) counts difference:";
1953                 }
1954                 break;
1955             case MAX_COUNTS_DIFFERENCE:
1956                 if ( treat_as_binary ) {
1957                     histogram_title = "counts difference:";
1958                 }
1959                 else {
1960                     histogram_title = "(maximal) counts difference:";
1961                 }
1962                 break;
1963             case DOMAIN_ID:
1964                 histogram_title = "score mean:";
1965                 break;
1966             case MIN:
1967                 histogram_title = "score minimum:";
1968                 break;
1969             case MAX:
1970                 histogram_title = "score maximum:";
1971                 break;
1972             case MAX_DIFFERENCE:
1973                 if ( treat_as_binary ) {
1974                     histogram_title = "difference:";
1975                 }
1976                 else {
1977                     histogram_title = "(maximal) difference:";
1978                 }
1979                 break;
1980             case MEAN:
1981                 histogram_title = "score mean:";
1982                 break;
1983             case SD:
1984                 histogram_title = "score standard deviation:";
1985                 break;
1986             case SPECIES_COUNT:
1987                 histogram_title = "species number:";
1988                 break;
1989             default:
1990                 throw new AssertionError( "Unknown sort field: " + sort_field );
1991         }
1992         for( final DomainSimilarity similarity : similarities ) {
1993             switch ( sort_field ) {
1994                 case ABS_MAX_COUNTS_DIFFERENCE:
1995                     stats.addValue( Math.abs( similarity.getMaximalDifferenceInCounts() ) );
1996                     break;
1997                 case MAX_COUNTS_DIFFERENCE:
1998                     stats.addValue( similarity.getMaximalDifferenceInCounts() );
1999                     break;
2000                 case DOMAIN_ID:
2001                     stats.addValue( similarity.getMeanSimilarityScore() );
2002                     break;
2003                 case MIN:
2004                     stats.addValue( similarity.getMinimalSimilarityScore() );
2005                     break;
2006                 case MAX:
2007                     stats.addValue( similarity.getMaximalSimilarityScore() );
2008                     break;
2009                 case MAX_DIFFERENCE:
2010                     stats.addValue( similarity.getMaximalDifference() );
2011                     break;
2012                 case MEAN:
2013                     stats.addValue( similarity.getMeanSimilarityScore() );
2014                     break;
2015                 case SD:
2016                     stats.addValue( similarity.getStandardDeviationOfSimilarityScore() );
2017                     break;
2018                 case SPECIES_COUNT:
2019                     stats.addValue( similarity.getSpecies().size() );
2020                     break;
2021                 default:
2022                     throw new AssertionError( "Unknown sort field: " + sort_field );
2023             }
2024         }
2025         //
2026         // final HistogramData[] hists = new HistogramData[ 1 ];
2027         //
2028         //
2029         // List<HistogramDataItem> data_items = new
2030         // ArrayList<HistogramDataItem>();
2031         // double[] values = stats.getDataAsDoubleArray();
2032         // for( int i = 0; i < values.length; i++ ) {
2033         // HistogramDataItem data_item = new BasicHistogramDataItem( "", values[
2034         // i ] );
2035         // data_items.add( data_item );
2036         // }
2037         //
2038         //
2039         // HistogramData hd0 = new HistogramData( "name",
2040         // data_items,
2041         // null, 20,
2042         // 40 );
2043         //
2044         //
2045         //
2046         //
2047         // hists[ 0 ] = hd0;
2048         //
2049         // final HistogramsFrame hf = new HistogramsFrame( hists );
2050         // hf.setVisible( true );
2051         //
2052         AsciiHistogram histo = null;
2053         if ( stats.getMin() < stats.getMin() ) {
2054             histo = new AsciiHistogram( stats, histogram_title );
2055         }
2056         if ( verbose ) {
2057             if ( histo != null ) {
2058                 System.out.println( histo.toStringBuffer( 20, '|', 40, 5 ) );
2059             }
2060             System.out.println();
2061             System.out.println( "N                   : " + stats.getN() );
2062             System.out.println( "Min                 : " + stats.getMin() );
2063             System.out.println( "Max                 : " + stats.getMax() );
2064             System.out.println( "Mean                : " + stats.arithmeticMean() );
2065             if ( stats.getN() > 1 ) {
2066                 System.out.println( "SD                  : " + stats.sampleStandardDeviation() );
2067             }
2068             else {
2069                 System.out.println( "SD                  : n/a" );
2070             }
2071             System.out.println( "Median              : " + stats.median() );
2072             if ( stats.getN() > 1 ) {
2073                 System.out.println( "Pearsonian skewness : " + stats.pearsonianSkewness() );
2074             }
2075             else {
2076                 System.out.println( "Pearsonian skewness : n/a" );
2077             }
2078         }
2079         switch ( print_option ) {
2080             case SIMPLE_TAB_DELIMITED:
2081                 break;
2082             case HTML:
2083                 w.write( "<html>" );
2084                 w.write( SurfacingConstants.NL );
2085                 addHtmlHead( w, "SURFACING :: " + html_title );
2086                 w.write( SurfacingConstants.NL );
2087                 w.write( "<body>" );
2088                 w.write( SurfacingConstants.NL );
2089                 w.write( html_desc.toString() );
2090                 w.write( SurfacingConstants.NL );
2091                 w.write( "<hr>" );
2092                 w.write( "<br>" );
2093                 w.write( SurfacingConstants.NL );
2094                 w.write( "<tt><pre>" );
2095                 w.write( SurfacingConstants.NL );
2096                 if ( histo != null ) {
2097                     w.write( histo.toStringBuffer( 20, '|', 40, 5 ).toString() );
2098                     w.write( SurfacingConstants.NL );
2099                 }
2100                 w.write( "</pre></tt>" );
2101                 w.write( SurfacingConstants.NL );
2102                 w.write( "<table>" );
2103                 w.write( SurfacingConstants.NL );
2104                 w.write( "<tr><td>N: </td><td>" + stats.getN() + "</td></tr>" );
2105                 w.write( SurfacingConstants.NL );
2106                 w.write( "<tr><td>Min: </td><td>" + stats.getMin() + "</td></tr>" );
2107                 w.write( SurfacingConstants.NL );
2108                 w.write( "<tr><td>Max: </td><td>" + stats.getMax() + "</td></tr>" );
2109                 w.write( SurfacingConstants.NL );
2110                 w.write( "<tr><td>Mean: </td><td>" + stats.arithmeticMean() + "</td></tr>" );
2111                 w.write( SurfacingConstants.NL );
2112                 if ( stats.getN() > 1 ) {
2113                     w.write( "<tr><td>SD: </td><td>" + stats.sampleStandardDeviation() + "</td></tr>" );
2114                 }
2115                 else {
2116                     w.write( "<tr><td>SD: </td><td>n/a</td></tr>" );
2117                 }
2118                 w.write( SurfacingConstants.NL );
2119                 w.write( "<tr><td>Median: </td><td>" + stats.median() + "</td></tr>" );
2120                 w.write( SurfacingConstants.NL );
2121                 if ( stats.getN() > 1 ) {
2122                     w.write( "<tr><td>Pearsonian skewness: </td><td>" + stats.pearsonianSkewness() + "</td></tr>" );
2123                 }
2124                 else {
2125                     w.write( "<tr><td>Pearsonian skewness: </td><td>n/a</td></tr>" );
2126                 }
2127                 w.write( SurfacingConstants.NL );
2128                 w.write( "</table>" );
2129                 w.write( SurfacingConstants.NL );
2130                 w.write( "<br>" );
2131                 w.write( SurfacingConstants.NL );
2132                 w.write( "<hr>" );
2133                 w.write( SurfacingConstants.NL );
2134                 w.write( "<br>" );
2135                 w.write( SurfacingConstants.NL );
2136                 w.write( "<table>" );
2137                 w.write( SurfacingConstants.NL );
2138                 break;
2139         }
2140         w.write( SurfacingConstants.NL );
2141         for( final DomainSimilarity similarity : similarities ) {
2142             if ( ( species_order != null ) && !species_order.isEmpty() ) {
2143                 ( ( PrintableDomainSimilarity ) similarity ).setSpeciesOrder( species_order );
2144             }
2145             w.write( similarity.toStringBuffer( print_option ).toString() );
2146             w.write( SurfacingConstants.NL );
2147         }
2148         switch ( print_option ) {
2149             case HTML:
2150                 w.write( SurfacingConstants.NL );
2151                 w.write( "</table>" );
2152                 w.write( SurfacingConstants.NL );
2153                 w.write( "</font>" );
2154                 w.write( SurfacingConstants.NL );
2155                 w.write( "</body>" );
2156                 w.write( SurfacingConstants.NL );
2157                 w.write( "</html>" );
2158                 w.write( SurfacingConstants.NL );
2159                 break;
2160         }
2161         w.flush();
2162         w.close();
2163         return stats;
2164     }
2165
2166     private static void writeDomainsToIndividualFilePerTreeNode( final Writer individual_files_writer,
2167                                                                  final String domain_0,
2168                                                                  final String domain_1 ) throws IOException {
2169         individual_files_writer.write( domain_0 );
2170         individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2171         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
2172             individual_files_writer.write( domain_1 );
2173             individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2174         }
2175     }
2176
2177     public static void writeMatrixToFile( final CharacterStateMatrix<?> matrix,
2178                                           final String filename,
2179                                           final Format format ) {
2180         final File outfile = new File( filename );
2181         checkForOutputFileWriteability( outfile );
2182         try {
2183             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
2184             matrix.toWriter( out, format );
2185             out.flush();
2186             out.close();
2187         }
2188         catch ( final IOException e ) {
2189             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2190         }
2191         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote matrix: \"" + filename + "\"" );
2192     }
2193
2194     public static void writeMatrixToFile( final File matrix_outfile, final List<DistanceMatrix> matrices ) {
2195         checkForOutputFileWriteability( matrix_outfile );
2196         try {
2197             final BufferedWriter out = new BufferedWriter( new FileWriter( matrix_outfile ) );
2198             for( final DistanceMatrix distance_matrix : matrices ) {
2199                 out.write( distance_matrix.toStringBuffer( DistanceMatrix.Format.PHYLIP ).toString() );
2200                 out.write( ForesterUtil.LINE_SEPARATOR );
2201                 out.flush();
2202             }
2203             out.close();
2204         }
2205         catch ( final IOException e ) {
2206             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2207         }
2208         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote distance matrices to \"" + matrix_outfile + "\"" );
2209     }
2210
2211     private static void writePfamsToFile( final String outfile_name, final SortedSet<String> pfams ) {
2212         try {
2213             final Writer writer = new BufferedWriter( new FileWriter( new File( outfile_name ) ) );
2214             for( final String pfam : pfams ) {
2215                 writer.write( pfam );
2216                 writer.write( ForesterUtil.LINE_SEPARATOR );
2217             }
2218             writer.close();
2219             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote " + pfams.size() + " pfams to [" + outfile_name
2220                     + "]" );
2221         }
2222         catch ( final IOException e ) {
2223             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
2224         }
2225     }
2226
2227     public static void writePhylogenyToFile( final Phylogeny phylogeny, final String filename ) {
2228         final PhylogenyWriter writer = new PhylogenyWriter();
2229         try {
2230             writer.toPhyloXML( new File( filename ), phylogeny, 1 );
2231         }
2232         catch ( final IOException e ) {
2233             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "failed to write phylogeny to \"" + filename + "\": "
2234                     + e );
2235         }
2236         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote phylogeny to \"" + filename + "\"" );
2237     }
2238
2239     public static void writeTaxonomyLinks( final Writer writer, final String species ) throws IOException {
2240         if ( ( species.length() > 1 ) && ( species.indexOf( '_' ) < 1 ) ) {
2241             final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( species );
2242             writer.write( " [" );
2243             if ( matcher.matches() ) {
2244                 writer.write( "<a href=\"" + SurfacingConstants.UNIPROT_LINK + species
2245                         + "\" target=\"taxonomy_window\">uniprot</a>" );
2246             }
2247             else {
2248                 writer.write( "<a href=\"" + SurfacingConstants.EOL_LINK + species
2249                         + "\" target=\"taxonomy_window\">eol</a>" );
2250                 writer.write( "|" );
2251                 writer.write( "<a href=\"" + SurfacingConstants.TOL_LINK + species
2252                         + "\" target=\"taxonomy_window\">tol</a>" );
2253             }
2254             writer.write( "]" );
2255         }
2256     }
2257
2258     private static void writeToNexus( final String outfile_name,
2259                                       final CharacterStateMatrix<BinaryStates> matrix,
2260                                       final Phylogeny phylogeny ) {
2261         if ( !( matrix instanceof BasicCharacterStateMatrix ) ) {
2262             throw new IllegalArgumentException( "can only write matrices of type [" + BasicCharacterStateMatrix.class
2263                     + "] to nexus" );
2264         }
2265         final BasicCharacterStateMatrix<BinaryStates> my_matrix = ( org.forester.evoinference.matrix.character.BasicCharacterStateMatrix<BinaryStates> ) matrix;
2266         final List<Phylogeny> phylogenies = new ArrayList<Phylogeny>( 1 );
2267         phylogenies.add( phylogeny );
2268         try {
2269             final BufferedWriter w = new BufferedWriter( new FileWriter( outfile_name ) );
2270             w.write( NexusConstants.NEXUS );
2271             w.write( ForesterUtil.LINE_SEPARATOR );
2272             my_matrix.writeNexusTaxaBlock( w );
2273             my_matrix.writeNexusBinaryChractersBlock( w );
2274             PhylogenyWriter.writeNexusTreesBlock( w, phylogenies );
2275             w.flush();
2276             w.close();
2277             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote Nexus file: \"" + outfile_name + "\"" );
2278         }
2279         catch ( final IOException e ) {
2280             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2281         }
2282     }
2283
2284     private static void writeToNexus( final String outfile_name,
2285                                       final DomainParsimonyCalculator domain_parsimony,
2286                                       final Phylogeny phylogeny ) {
2287         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAINS,
2288                       domain_parsimony.createMatrixOfDomainPresenceOrAbsence(),
2289                       phylogeny );
2290         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAIN_COMBINATIONS,
2291                       domain_parsimony.createMatrixOfBinaryDomainCombinationPresenceOrAbsence(),
2292                       phylogeny );
2293     }
2294
2295     public static void domainsPerProteinsStatistics( final String genome,
2296                                                      final List<Protein> protein_list,
2297                                                      final DescriptiveStatistics all_genomes_domains_per_potein_stats,
2298                                                      final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
2299                                                      final SortedSet<String> domains_which_are_always_single,
2300                                                      final SortedSet<String> domains_which_are_sometimes_single_sometimes_not,
2301                                                      final SortedSet<String> domains_which_never_single,
2302                                                      final Writer writer ) {
2303         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
2304         for( final Protein protein : protein_list ) {
2305             final int domains = protein.getNumberOfProteinDomains();
2306             //System.out.println( domains );
2307             stats.addValue( domains );
2308             all_genomes_domains_per_potein_stats.addValue( domains );
2309             if ( !all_genomes_domains_per_potein_histo.containsKey( domains ) ) {
2310                 all_genomes_domains_per_potein_histo.put( domains, 1 );
2311             }
2312             else {
2313                 all_genomes_domains_per_potein_histo.put( domains,
2314                                                           1 + all_genomes_domains_per_potein_histo.get( domains ) );
2315             }
2316             if ( domains == 1 ) {
2317                 final String domain = protein.getProteinDomain( 0 ).getDomainId().getId();
2318                 if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
2319                     if ( domains_which_never_single.contains( domain ) ) {
2320                         domains_which_never_single.remove( domain );
2321                         domains_which_are_sometimes_single_sometimes_not.add( domain );
2322                     }
2323                     else {
2324                         domains_which_are_always_single.add( domain );
2325                     }
2326                 }
2327             }
2328             else if ( domains > 1 ) {
2329                 for( final Domain d : protein.getProteinDomains() ) {
2330                     final String domain = d.getDomainId().getId();
2331                     // System.out.println( domain );
2332                     if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
2333                         if ( domains_which_are_always_single.contains( domain ) ) {
2334                             domains_which_are_always_single.remove( domain );
2335                             domains_which_are_sometimes_single_sometimes_not.add( domain );
2336                         }
2337                         else {
2338                             domains_which_never_single.add( domain );
2339                         }
2340                     }
2341                 }
2342             }
2343         }
2344         try {
2345             writer.write( genome );
2346             writer.write( "\t" );
2347             writer.write( stats.arithmeticMean() + "" );
2348             writer.write( "\t" );
2349             writer.write( stats.sampleStandardDeviation() + "" );
2350             writer.write( "\t" );
2351             writer.write( stats.median() + "" );
2352             writer.write( "\t" );
2353             writer.write( stats.getN() + "" );
2354             writer.write( "\t" );
2355             writer.write( stats.getMin() + "" );
2356             writer.write( "\t" );
2357             writer.write( stats.getMax() + "" );
2358             writer.write( "\n" );
2359         }
2360         catch ( final IOException e ) {
2361             e.printStackTrace();
2362         }
2363     }
2364 }