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.PhylogenyNodeI.NH_CONVERSION_SUPPORT_VALUE_STYLE;
  71 import org.forester.phylogeny.data.BinaryCharacters;
  72 import org.forester.phylogeny.data.Confidence;
  73 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  74 import org.forester.surfacing.DomainSimilarityCalculator.Detailedness;
  75 import org.forester.surfacing.DomainSimilarityCalculator.GoAnnotationOutput;
  76 import org.forester.surfacing.GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder;
  77 import org.forester.util.AsciiHistogram;
  78 import org.forester.util.BasicDescriptiveStatistics;
  79 import org.forester.util.BasicTable;
  80 import org.forester.util.BasicTableParser;
  81 import org.forester.util.DescriptiveStatistics;
  82 import org.forester.util.ForesterUtil;
  83
  84 public final class SurfacingUtil {
  85
  86     private final static NumberFormat       FORMATTER                        = new DecimalFormat( "0.0E0" );
  87     private final static NumberFormat       FORMATTER_3                      = new DecimalFormat( "0.000" );
  88     private static final Comparator<Domain> ASCENDING_CONFIDENCE_VALUE_ORDER = new Comparator<Domain>() {
  89
  90                                                                                  @Override
  91                                                                                  public int compare( final Domain d1,
  92                                                                                                      final Domain d2 ) {
  93                                                                                      if ( d1.getPerSequenceEvalue() < d2
  94                                                                                              .getPerSequenceEvalue() ) {
  95                                                                                          return -1;
  96                                                                                      }
  97                                                                                      else if ( d1
  98                                                                                              .getPerSequenceEvalue() > d2
  99                                                                                              .getPerSequenceEvalue() ) {
 100                                                                                          return 1;
 101                                                                                      }
 102                                                                                      else {
 103                                                                                          return d1.compareTo( d2 );
 104                                                                                      }
 105                                                                                  }
 106                                                                              };
 107     public final static Pattern             PATTERN_SP_STYLE_TAXONOMY        = Pattern.compile( "^[A-Z0-9]{3,5}$" );
 108
 109     private SurfacingUtil() {
 110         // Hidden constructor.
 111     }
 112
 113     public static void addAllBinaryDomainCombinationToSet( final GenomeWideCombinableDomains genome,
 114                                                            final SortedSet<BinaryDomainCombination> binary_domain_combinations ) {
 115         final SortedMap<DomainId, CombinableDomains> all_cd = genome.getAllCombinableDomainsIds();
 116         for( final DomainId domain_id : all_cd.keySet() ) {
 117             binary_domain_combinations.addAll( all_cd.get( domain_id ).toBinaryDomainCombinations() );
 118         }
 119     }
 120
 121     public static void addAllDomainIdsToSet( final GenomeWideCombinableDomains genome,
 122                                              final SortedSet<DomainId> domain_ids ) {
 123         final SortedSet<DomainId> domains = genome.getAllDomainIds();
 124         for( final DomainId domain : domains ) {
 125             domain_ids.add( domain );
 126         }
 127     }
 128
 129     public static void addHtmlHead( final Writer w, final String title ) throws IOException {
 130         w.write( SurfacingConstants.NL );
 131         w.write( "<head>" );
 132         w.write( "<title>" );
 133         w.write( title );
 134         w.write( "</title>" );
 135         w.write( SurfacingConstants.NL );
 136         w.write( "<style>" );
 137         w.write( SurfacingConstants.NL );
 138         w.write( "a:visited { color : #6633FF; text-decoration : none; }" );
 139         w.write( SurfacingConstants.NL );
 140         w.write( "a:link { color : #6633FF; text-decoration : none; }" );
 141         w.write( SurfacingConstants.NL );
 142         w.write( "a:active { color : #99FF00; text-decoration : none; }" );
 143         w.write( SurfacingConstants.NL );
 144         w.write( "a:hover { color : #FFFFFF; background-color : #99FF00; text-decoration : none; }" );
 145         w.write( SurfacingConstants.NL );
 146         w.write( "td { text-align: left; vertical-align: top; font-family: Verdana, Arial, Helvetica; font-size: 8pt}" );
 147         w.write( SurfacingConstants.NL );
 148         w.write( "h1 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 18pt; font-weight: bold }" );
 149         w.write( SurfacingConstants.NL );
 150         w.write( "h2 { color : #0000FF; font-family: Verdana, Arial, Helvetica; font-size: 16pt; font-weight: bold }" );
 151         w.write( SurfacingConstants.NL );
 152         w.write( "</style>" );
 153         w.write( SurfacingConstants.NL );
 154         w.write( "</head>" );
 155         w.write( SurfacingConstants.NL );
 156     }
 157
 158     public static DescriptiveStatistics calculateDescriptiveStatisticsForMeanValues( final Set<DomainSimilarity> similarities ) {
 159         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 160         for( final DomainSimilarity similarity : similarities ) {
 161             stats.addValue( similarity.getMeanSimilarityScore() );
 162         }
 163         return stats;
 164     }
 165
 166     private static void calculateIndependentDomainCombinationGains( final Phylogeny local_phylogeny_l,
 167                                                                     final String outfilename_for_counts,
 168                                                                     final String outfilename_for_dc,
 169                                                                     final String outfilename_for_dc_for_go_mapping,
 170                                                                     final String outfilename_for_dc_for_go_mapping_unique,
 171                                                                     final String outfilename_for_rank_counts,
 172                                                                     final String outfilename_for_ancestor_species_counts ) {
 173         try {
 174             final BufferedWriter out_counts = new BufferedWriter( new FileWriter( outfilename_for_counts ) );
 175             final BufferedWriter out_dc = new BufferedWriter( new FileWriter( outfilename_for_dc ) );
 176             final BufferedWriter out_dc_for_go_mapping = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping ) );
 177             final BufferedWriter out_dc_for_go_mapping_unique = new BufferedWriter( new FileWriter( outfilename_for_dc_for_go_mapping_unique ) );
 178             final SortedMap<String, Integer> dc_gain_counts = new TreeMap<String, Integer>();
 179             for( final PhylogenyNodeIterator it = local_phylogeny_l.iteratorPostorder(); it.hasNext(); ) {
 180                 final PhylogenyNode n = it.next();
 181                 final Set<String> gained_dc = n.getNodeData().getBinaryCharacters().getGainedCharacters();
 182                 for( final String dc : gained_dc ) {
 183                     if ( dc_gain_counts.containsKey( dc ) ) {
 184                         dc_gain_counts.put( dc, dc_gain_counts.get( dc ) + 1 );
 185                     }
 186                     else {
 187                         dc_gain_counts.put( dc, 1 );
 188                     }
 189                 }
 190             }
 191             final SortedMap<Integer, Integer> histogram = new TreeMap<Integer, Integer>();
 192             final SortedMap<Integer, StringBuilder> domain_lists = new TreeMap<Integer, StringBuilder>();
 193             final SortedMap<Integer, PriorityQueue<String>> domain_lists_go = new TreeMap<Integer, PriorityQueue<String>>();
 194             final SortedMap<Integer, SortedSet<String>> domain_lists_go_unique = new TreeMap<Integer, SortedSet<String>>();
 195             final Set<String> dcs = dc_gain_counts.keySet();
 196             final SortedSet<String> more_than_once = new TreeSet<String>();
 197             for( final String dc : dcs ) {
 198                 final int count = dc_gain_counts.get( dc );
 199                 if ( histogram.containsKey( count ) ) {
 200                     histogram.put( count, histogram.get( count ) + 1 );
 201                     domain_lists.get( count ).append( ", " + dc );
 202                     domain_lists_go.get( count ).addAll( splitDomainCombination( dc ) );
 203                     domain_lists_go_unique.get( count ).addAll( splitDomainCombination( dc ) );
 204                 }
 205                 else {
 206                     histogram.put( count, 1 );
 207                     domain_lists.put( count, new StringBuilder( dc ) );
 208                     final PriorityQueue<String> q = new PriorityQueue<String>();
 209                     q.addAll( splitDomainCombination( dc ) );
 210                     domain_lists_go.put( count, q );
 211                     final SortedSet<String> set = new TreeSet<String>();
 212                     set.addAll( splitDomainCombination( dc ) );
 213                     domain_lists_go_unique.put( count, set );
 214                 }
 215                 if ( count > 1 ) {
 216                     more_than_once.add( dc );
 217                 }
 218             }
 219             final Set<Integer> histogram_keys = histogram.keySet();
 220             for( final Integer histogram_key : histogram_keys ) {
 221                 final int count = histogram.get( histogram_key );
 222                 final StringBuilder dc = domain_lists.get( histogram_key );
 223                 out_counts.write( histogram_key + "\t" + count + ForesterUtil.LINE_SEPARATOR );
 224                 out_dc.write( histogram_key + "\t" + dc + ForesterUtil.LINE_SEPARATOR );
 225                 out_dc_for_go_mapping.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
 226                 final Object[] sorted = domain_lists_go.get( histogram_key ).toArray();
 227                 Arrays.sort( sorted );
 228                 for( final Object domain : sorted ) {
 229                     out_dc_for_go_mapping.write( domain + ForesterUtil.LINE_SEPARATOR );
 230                 }
 231                 out_dc_for_go_mapping_unique.write( "#" + histogram_key + ForesterUtil.LINE_SEPARATOR );
 232                 for( final String domain : domain_lists_go_unique.get( histogram_key ) ) {
 233                     out_dc_for_go_mapping_unique.write( domain + ForesterUtil.LINE_SEPARATOR );
 234                 }
 235             }
 236             out_counts.close();
 237             out_dc.close();
 238             out_dc_for_go_mapping.close();
 239             out_dc_for_go_mapping_unique.close();
 240             //
 241             final SortedMap<String, Integer> lca_rank_counts = new TreeMap<String, Integer>();
 242             final SortedMap<String, Integer> lca_ancestor_species_counts = new TreeMap<String, Integer>();
 243             for( final String dc : more_than_once ) {
 244                 final List<PhylogenyNode> nodes = new ArrayList<PhylogenyNode>();
 245                 for( final PhylogenyNodeIterator it = local_phylogeny_l.iteratorExternalForward(); it.hasNext(); ) {
 246                     final PhylogenyNode n = it.next();
 247                     if ( n.getNodeData().getBinaryCharacters().getGainedCharacters().contains( dc ) ) {
 248                         nodes.add( n );
 249                     }
 250                 }
 251                 for( int i = 0; i < nodes.size() - 1; ++i ) {
 252                     for( int j = i + 1; j < nodes.size(); ++j ) {
 253                         final PhylogenyNode lca = PhylogenyMethods.getInstance().obtainLCA( nodes.get( i ),
 254                                                                                             nodes.get( j ) );
 255                         String rank = "unknown";
 256                         if ( lca.getNodeData().isHasTaxonomy()
 257                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getRank() ) ) {
 258                             rank = lca.getNodeData().getTaxonomy().getRank();
 259                         }
 260                         addToCountMap( lca_rank_counts, rank );
 261                         String lca_species;
 262                         if ( lca.getNodeData().isHasTaxonomy()
 263                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getScientificName() ) ) {
 264                             lca_species = lca.getNodeData().getTaxonomy().getScientificName();
 265                         }
 266                         else if ( lca.getNodeData().isHasTaxonomy()
 267                                 && !ForesterUtil.isEmpty( lca.getNodeData().getTaxonomy().getCommonName() ) ) {
 268                             lca_species = lca.getNodeData().getTaxonomy().getCommonName();
 269                         }
 270                         else {
 271                             lca_species = lca.getName();
 272                         }
 273                         addToCountMap( lca_ancestor_species_counts, lca_species );
 274                     }
 275                 }
 276             }
 277             final BufferedWriter out_for_rank_counts = new BufferedWriter( new FileWriter( outfilename_for_rank_counts ) );
 278             final BufferedWriter out_for_ancestor_species_counts = new BufferedWriter( new FileWriter( outfilename_for_ancestor_species_counts ) );
 279             ForesterUtil.map2writer( out_for_rank_counts, lca_rank_counts, "\t", ForesterUtil.LINE_SEPARATOR );
 280             ForesterUtil.map2writer( out_for_ancestor_species_counts,
 281                                      lca_ancestor_species_counts,
 282                                      "\t",
 283                                      ForesterUtil.LINE_SEPARATOR );
 284             out_for_rank_counts.close();
 285             out_for_ancestor_species_counts.close();
 286         }
 287         catch ( final IOException e ) {
 288             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
 289         }
 290         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch counts to ["
 291                 + outfilename_for_counts + "]" );
 292         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote independent domain combination gains fitch lists to ["
 293                 + outfilename_for_dc + "]" );
 294         ForesterUtil.programMessage( surfacing.PRG_NAME,
 295                                      "Wrote independent domain combination gains fitch lists to (for GO mapping) ["
 296                                              + outfilename_for_dc_for_go_mapping + "]" );
 297         ForesterUtil.programMessage( surfacing.PRG_NAME,
 298                                      "Wrote independent domain combination gains fitch lists to (for GO mapping, unique) ["
 299                                              + outfilename_for_dc_for_go_mapping_unique + "]" );
 300     }
 301
 302     private final static void addToCountMap( final Map<String, Integer> map, final String s ) {
 303         if ( map.containsKey( s ) ) {
 304             map.put( s, map.get( s ) + 1 );
 305         }
 306         else {
 307             map.put( s, 1 );
 308         }
 309     }
 310
 311     public static int calculateOverlap( final Domain domain, final List<Boolean> covered_positions ) {
 312         int overlap_count = 0;
 313         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
 314             if ( ( i < covered_positions.size() ) && ( covered_positions.get( i ) == true ) ) {
 315                 ++overlap_count;
 316             }
 317         }
 318         return overlap_count;
 319     }
 320
 321     public static void checkForOutputFileWriteability( final File outfile ) {
 322         final String error = ForesterUtil.isWritableFile( outfile );
 323         if ( !ForesterUtil.isEmpty( error ) ) {
 324             ForesterUtil.fatalError( surfacing.PRG_NAME, error );
 325         }
 326     }
 327
 328     private static SortedSet<String> collectAllDomainsChangedOnSubtree( final PhylogenyNode subtree_root,
 329                                                                         final boolean get_gains ) {
 330         final SortedSet<String> domains = new TreeSet<String>();
 331         for( final PhylogenyNode descendant : PhylogenyMethods.getAllDescendants( subtree_root ) ) {
 332             final BinaryCharacters chars = descendant.getNodeData().getBinaryCharacters();
 333             if ( get_gains ) {
 334                 domains.addAll( chars.getGainedCharacters() );
 335             }
 336             else {
 337                 domains.addAll( chars.getLostCharacters() );
 338             }
 339         }
 340         return domains;
 341     }
 342
 343     public static void collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
 344                                                                                            final BinaryDomainCombination.DomainCombinationType dc_type,
 345                                                                                            final List<BinaryDomainCombination> all_binary_domains_combination_gained,
 346                                                                                            final boolean get_gains ) {
 347         final SortedSet<String> sorted_ids = new TreeSet<String>();
 348         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
 349             sorted_ids.add( matrix.getIdentifier( i ) );
 350         }
 351         for( final String id : sorted_ids ) {
 352             for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
 353                 if ( ( get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) )
 354                         || ( !get_gains && ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.LOSS ) ) ) {
 355                     if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED_ADJACTANT ) {
 356                         all_binary_domains_combination_gained.add( AdjactantDirectedBinaryDomainCombination
 357                                 .createInstance( matrix.getCharacter( c ) ) );
 358                     }
 359                     else if ( dc_type == BinaryDomainCombination.DomainCombinationType.DIRECTED ) {
 360                         all_binary_domains_combination_gained.add( DirectedBinaryDomainCombination
 361                                 .createInstance( matrix.getCharacter( c ) ) );
 362                     }
 363                     else {
 364                         all_binary_domains_combination_gained.add( BasicBinaryDomainCombination.createInstance( matrix
 365                                 .getCharacter( c ) ) );
 366                     }
 367                 }
 368             }
 369         }
 370     }
 371
 372     private static File createBaseDirForPerNodeDomainFiles( final String base_dir,
 373                                                             final boolean domain_combinations,
 374                                                             final CharacterStateMatrix.GainLossStates state,
 375                                                             final String outfile ) {
 376         File per_node_go_mapped_domain_gain_loss_files_base_dir = new File( new File( outfile ).getParent()
 377                 + ForesterUtil.FILE_SEPARATOR + base_dir );
 378         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 379             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 380         }
 381         if ( domain_combinations ) {
 382             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 383                     + ForesterUtil.FILE_SEPARATOR + "DC" );
 384         }
 385         else {
 386             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 387                     + ForesterUtil.FILE_SEPARATOR + "DOMAINS" );
 388         }
 389         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 390             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 391         }
 392         if ( state == GainLossStates.GAIN ) {
 393             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 394                     + ForesterUtil.FILE_SEPARATOR + "GAINS" );
 395         }
 396         else if ( state == GainLossStates.LOSS ) {
 397             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 398                     + ForesterUtil.FILE_SEPARATOR + "LOSSES" );
 399         }
 400         else {
 401             per_node_go_mapped_domain_gain_loss_files_base_dir = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
 402                     + ForesterUtil.FILE_SEPARATOR + "PRESENT" );
 403         }
 404         if ( !per_node_go_mapped_domain_gain_loss_files_base_dir.exists() ) {
 405             per_node_go_mapped_domain_gain_loss_files_base_dir.mkdir();
 406         }
 407         return per_node_go_mapped_domain_gain_loss_files_base_dir;
 408     }
 409
 410     public static Map<DomainId, List<GoId>> createDomainIdToGoIdMap( final List<PfamToGoMapping> pfam_to_go_mappings ) {
 411         final Map<DomainId, List<GoId>> domain_id_to_go_ids_map = new HashMap<DomainId, List<GoId>>( pfam_to_go_mappings
 412                 .size() );
 413         for( final PfamToGoMapping pfam_to_go : pfam_to_go_mappings ) {
 414             if ( !domain_id_to_go_ids_map.containsKey( pfam_to_go.getKey() ) ) {
 415                 domain_id_to_go_ids_map.put( pfam_to_go.getKey(), new ArrayList<GoId>() );
 416             }
 417             domain_id_to_go_ids_map.get( pfam_to_go.getKey() ).add( pfam_to_go.getValue() );
 418         }
 419         return domain_id_to_go_ids_map;
 420     }
 421
 422     public static Map<DomainId, Set<String>> createDomainIdToSecondaryFeaturesMap( final File secondary_features_map_file )
 423             throws IOException {
 424         final BasicTable<String> primary_table = BasicTableParser.parse( secondary_features_map_file, "\t" );
 425         final Map<DomainId, Set<String>> map = new TreeMap<DomainId, Set<String>>();
 426         for( int r = 0; r < primary_table.getNumberOfRows(); ++r ) {
 427             final DomainId domain_id = new DomainId( primary_table.getValue( 0, r ) );
 428             if ( !map.containsKey( domain_id ) ) {
 429                 map.put( domain_id, new HashSet<String>() );
 430             }
 431             map.get( domain_id ).add( primary_table.getValue( 1, r ) );
 432         }
 433         return map;
 434     }
 435
 436     public static Phylogeny createNjTreeBasedOnMatrixToFile( final File nj_tree_outfile, final DistanceMatrix distance ) {
 437         checkForOutputFileWriteability( nj_tree_outfile );
 438         final NeighborJoining nj = NeighborJoining.createInstance();
 439         final Phylogeny phylogeny = nj.execute( distance );
 440         phylogeny.setName( nj_tree_outfile.getName() );
 441         writePhylogenyToFile( phylogeny, nj_tree_outfile.toString() );
 442         return phylogeny;
 443     }
 444
 445     private static SortedSet<BinaryDomainCombination> createSetOfAllBinaryDomainCombinationsPerGenome( final GenomeWideCombinableDomains gwcd ) {
 446         final SortedMap<DomainId, CombinableDomains> cds = gwcd.getAllCombinableDomainsIds();
 447         final SortedSet<BinaryDomainCombination> binary_combinations = new TreeSet<BinaryDomainCombination>();
 448         for( final DomainId domain_id : cds.keySet() ) {
 449             final CombinableDomains cd = cds.get( domain_id );
 450             binary_combinations.addAll( cd.toBinaryDomainCombinations() );
 451         }
 452         return binary_combinations;
 453     }
 454
 455     public static void decoratePrintableDomainSimilarities( final SortedSet<DomainSimilarity> domain_similarities,
 456                                                             final Detailedness detailedness,
 457                                                             final GoAnnotationOutput go_annotation_output,
 458                                                             final Map<GoId, GoTerm> go_id_to_term_map,
 459                                                             final GoNameSpace go_namespace_limit ) {
 460         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || go_id_to_term_map.isEmpty() ) ) {
 461             throw new IllegalArgumentException( "attempt to use a GO namespace limit without a GO id to term map" );
 462         }
 463         for( final DomainSimilarity domain_similarity : domain_similarities ) {
 464             if ( domain_similarity instanceof PrintableDomainSimilarity ) {
 465                 final PrintableDomainSimilarity printable_domain_similarity = ( PrintableDomainSimilarity ) domain_similarity;
 466                 printable_domain_similarity.setDetailedness( detailedness );
 467                 printable_domain_similarity.setGoAnnotationOutput( go_annotation_output );
 468                 printable_domain_similarity.setGoIdToTermMap( go_id_to_term_map );
 469                 printable_domain_similarity.setGoNamespaceLimit( go_namespace_limit );
 470             }
 471         }
 472     }
 473
 474     public static void executeDomainLengthAnalysis( final String[][] input_file_properties,
 475                                                     final int number_of_genomes,
 476                                                     final DomainLengthsTable domain_lengths_table,
 477                                                     final File outfile ) throws IOException {
 478         final DecimalFormat df = new DecimalFormat( "#.00" );
 479         checkForOutputFileWriteability( outfile );
 480         final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
 481         out.write( "MEAN BASED STATISTICS PER SPECIES" );
 482         out.write( ForesterUtil.LINE_SEPARATOR );
 483         out.write( domain_lengths_table.createMeanBasedStatisticsPerSpeciesTable().toString() );
 484         out.write( ForesterUtil.LINE_SEPARATOR );
 485         out.write( ForesterUtil.LINE_SEPARATOR );
 486         final List<DomainLengths> domain_lengths_list = domain_lengths_table.getDomainLengthsList();
 487         out.write( "OUTLIER SPECIES PER DOMAIN (Z>=1.5)" );
 488         out.write( ForesterUtil.LINE_SEPARATOR );
 489         for( final DomainLengths domain_lengths : domain_lengths_list ) {
 490             final List<Species> species_list = domain_lengths.getMeanBasedOutlierSpecies( 1.5 );
 491             if ( species_list.size() > 0 ) {
 492                 out.write( domain_lengths.getDomainId() + "\t" );
 493                 for( final Species species : species_list ) {
 494                     out.write( species + "\t" );
 495                 }
 496                 out.write( ForesterUtil.LINE_SEPARATOR );
 497                 // DescriptiveStatistics stats_for_domain = domain_lengths
 498                 //         .calculateMeanBasedStatistics();
 499                 //AsciiHistogram histo = new AsciiHistogram( stats_for_domain );
 500                 //System.out.println( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 501             }
 502         }
 503         out.write( ForesterUtil.LINE_SEPARATOR );
 504         out.write( ForesterUtil.LINE_SEPARATOR );
 505         out.write( "OUTLIER SPECIES (Z 1.0)" );
 506         out.write( ForesterUtil.LINE_SEPARATOR );
 507         final DescriptiveStatistics stats_for_all_species = domain_lengths_table
 508                 .calculateMeanBasedStatisticsForAllSpecies();
 509         out.write( stats_for_all_species.asSummary() );
 510         out.write( ForesterUtil.LINE_SEPARATOR );
 511         final AsciiHistogram histo = new AsciiHistogram( stats_for_all_species );
 512         out.write( histo.toStringBuffer( 40, '=', 60, 4 ).toString() );
 513         out.write( ForesterUtil.LINE_SEPARATOR );
 514         final double population_sd = stats_for_all_species.sampleStandardDeviation();
 515         final double population_mean = stats_for_all_species.arithmeticMean();
 516         for( final Species species : domain_lengths_table.getSpecies() ) {
 517             final double x = domain_lengths_table.calculateMeanBasedStatisticsForSpecies( species ).arithmeticMean();
 518             final double z = ( x - population_mean ) / population_sd;
 519             out.write( species + "\t" + z );
 520             out.write( ForesterUtil.LINE_SEPARATOR );
 521         }
 522         out.write( ForesterUtil.LINE_SEPARATOR );
 523         for( final Species species : domain_lengths_table.getSpecies() ) {
 524             final DescriptiveStatistics stats_for_species = domain_lengths_table
 525                     .calculateMeanBasedStatisticsForSpecies( species );
 526             final double x = stats_for_species.arithmeticMean();
 527             final double z = ( x - population_mean ) / population_sd;
 528             if ( ( z <= -1.0 ) || ( z >= 1.0 ) ) {
 529                 out.write( species + "\t" + df.format( z ) + "\t" + stats_for_species.asSummary() );
 530                 out.write( ForesterUtil.LINE_SEPARATOR );
 531             }
 532         }
 533         out.close();
 534         //        final List<HistogramData> histogram_datas = new ArrayList<HistogramData>();
 535         //        for( int i = 0; i < number_of_genomes; ++i ) {
 536         //            final Species species = new BasicSpecies( input_file_properties[ i ][ 0 ] );
 537         //            histogram_datas
 538         //                    .add( new HistogramData( species.toString(), domain_lengths_table
 539         //                            .calculateMeanBasedStatisticsForSpecies( species )
 540         //                            .getDataAsDoubleArray(), 5, 600, null, 60 ) );
 541         //        }
 542         //        final HistogramsFrame hf = new HistogramsFrame( histogram_datas );
 543         //        hf.setVisible( true );
 544         System.gc();
 545     }
 546
 547     /**
 548      *
 549      * @param all_binary_domains_combination_lost_fitch
 550      * @param consider_directedness_and_adjacency_for_bin_combinations
 551      * @param all_binary_domains_combination_gained if null ignored, otherwise this is to list all binary domain combinations
 552      * which were gained under unweighted (Fitch) parsimony.
 553      */
 554     public static void executeParsimonyAnalysis( final long random_number_seed_for_fitch_parsimony,
 555                                                  final boolean radomize_fitch_parsimony,
 556                                                  final String outfile_name,
 557                                                  final DomainParsimonyCalculator domain_parsimony,
 558                                                  final Phylogeny phylogeny,
 559                                                  final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 560                                                  final Map<GoId, GoTerm> go_id_to_term_map,
 561                                                  final GoNameSpace go_namespace_limit,
 562                                                  final String parameters_str,
 563                                                  final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
 564                                                  final SortedSet<DomainId> positive_filter,
 565                                                  final boolean output_binary_domain_combinations_for_graphs,
 566                                                  final List<BinaryDomainCombination> all_binary_domains_combination_gained_fitch,
 567                                                  final List<BinaryDomainCombination> all_binary_domains_combination_lost_fitch,
 568                                                  final BinaryDomainCombination.DomainCombinationType dc_type ) {
 569         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 570         final String date_time = ForesterUtil.getCurrentDateTime();
 571         final SortedSet<String> all_pfams_encountered = new TreeSet<String>();
 572         final SortedSet<String> all_pfams_gained_as_domains = new TreeSet<String>();
 573         final SortedSet<String> all_pfams_lost_as_domains = new TreeSet<String>();
 574         final SortedSet<String> all_pfams_gained_as_dom_combinations = new TreeSet<String>();
 575         final SortedSet<String> all_pfams_lost_as_dom_combinations = new TreeSet<String>();
 576         writeToNexus( outfile_name, domain_parsimony, phylogeny );
 577         // DOLLO DOMAINS
 578         // -------------
 579         Phylogeny local_phylogeny_l = phylogeny.copy();
 580         if ( ( positive_filter != null ) && ( positive_filter.size() > 0 ) ) {
 581             domain_parsimony.executeDolloParsimonyOnDomainPresence( positive_filter );
 582         }
 583         else {
 584             domain_parsimony.executeDolloParsimonyOnDomainPresence();
 585         }
 586         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 587                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 588         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 589                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS, Format.FORESTER );
 590         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 591                                                            CharacterStateMatrix.GainLossStates.GAIN,
 592                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_D,
 593                                                            sep,
 594                                                            ForesterUtil.LINE_SEPARATOR,
 595                                                            null );
 596         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 597                                                            CharacterStateMatrix.GainLossStates.LOSS,
 598                                                            outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_D,
 599                                                            sep,
 600                                                            ForesterUtil.LINE_SEPARATOR,
 601                                                            null );
 602         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 603                 + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_D, sep, ForesterUtil.LINE_SEPARATOR, null );
 604         //HTML:
 605         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 606                                        go_id_to_term_map,
 607                                        go_namespace_limit,
 608                                        false,
 609                                        domain_parsimony.getGainLossMatrix(),
 610                                        CharacterStateMatrix.GainLossStates.GAIN,
 611                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D,
 612                                        sep,
 613                                        ForesterUtil.LINE_SEPARATOR,
 614                                        "Dollo Parsimony | Gains | Domains",
 615                                        "+",
 616                                        domain_id_to_secondary_features_maps,
 617                                        all_pfams_encountered,
 618                                        all_pfams_gained_as_domains,
 619                                        "_dollo_gains_d" );
 620         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 621                                        go_id_to_term_map,
 622                                        go_namespace_limit,
 623                                        false,
 624                                        domain_parsimony.getGainLossMatrix(),
 625                                        CharacterStateMatrix.GainLossStates.LOSS,
 626                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D,
 627                                        sep,
 628                                        ForesterUtil.LINE_SEPARATOR,
 629                                        "Dollo Parsimony | Losses | Domains",
 630                                        "-",
 631                                        domain_id_to_secondary_features_maps,
 632                                        all_pfams_encountered,
 633                                        all_pfams_lost_as_domains,
 634                                        "_dollo_losses_d" );
 635         writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 636                                        go_id_to_term_map,
 637                                        go_namespace_limit,
 638                                        false,
 639                                        domain_parsimony.getGainLossMatrix(),
 640                                        null,
 641                                        outfile_name + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D,
 642                                        sep,
 643                                        ForesterUtil.LINE_SEPARATOR,
 644                                        "Dollo Parsimony | Present | Domains",
 645                                        "",
 646                                        domain_id_to_secondary_features_maps,
 647                                        all_pfams_encountered,
 648                                        null,
 649                                        "_dollo_present_d" );
 650         preparePhylogeny( local_phylogeny_l,
 651                           domain_parsimony,
 652                           date_time,
 653                           "Dollo parsimony on domain presence/absence",
 654                           "dollo_on_domains_" + outfile_name,
 655                           parameters_str );
 656         SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 657                 + surfacing.DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 658         try {
 659             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, true, outfile_name, "_dollo_all_gains_d" );
 660             writeAllDomainsChangedOnAllSubtrees( local_phylogeny_l, false, outfile_name, "_dollo_all_losses_d" );
 661         }
 662         catch ( final IOException e ) {
 663             e.printStackTrace();
 664             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
 665         }
 666         if ( domain_parsimony.calculateNumberOfBinaryDomainCombination() > 0 ) {
 667             // FITCH DOMAIN COMBINATIONS
 668             // -------------------------
 669             local_phylogeny_l = phylogeny.copy();
 670             String randomization = "no";
 671             if ( radomize_fitch_parsimony ) {
 672                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( random_number_seed_for_fitch_parsimony );
 673                 randomization = "yes, seed = " + random_number_seed_for_fitch_parsimony;
 674             }
 675             else {
 676                 domain_parsimony.executeFitchParsimonyOnBinaryDomainCombintion( true );
 677             }
 678             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(), outfile_name
 679                     + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 680             SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(), outfile_name
 681                     + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS, Format.FORESTER );
 682             SurfacingUtil
 683                     .writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 684                                                           CharacterStateMatrix.GainLossStates.GAIN,
 685                                                           outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_BC,
 686                                                           sep,
 687                                                           ForesterUtil.LINE_SEPARATOR,
 688                                                           null );
 689             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 690                                                                CharacterStateMatrix.GainLossStates.LOSS,
 691                                                                outfile_name
 692                                                                        + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_BC,
 693                                                                sep,
 694                                                                ForesterUtil.LINE_SEPARATOR,
 695                                                                null );
 696             SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 697                     + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC, sep, ForesterUtil.LINE_SEPARATOR, null );
 698             if ( all_binary_domains_combination_gained_fitch != null ) {
 699                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 700                                                                                     dc_type,
 701                                                                                     all_binary_domains_combination_gained_fitch,
 702                                                                                     true );
 703             }
 704             if ( all_binary_domains_combination_lost_fitch != null ) {
 705                 collectChangedDomainCombinationsFromBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 706                                                                                     dc_type,
 707                                                                                     all_binary_domains_combination_lost_fitch,
 708                                                                                     false );
 709             }
 710             if ( output_binary_domain_combinations_for_graphs ) {
 711                 SurfacingUtil
 712                         .writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( domain_parsimony
 713                                                                                                            .getGainLossMatrix(),
 714                                                                                                    null,
 715                                                                                                    outfile_name
 716                                                                                                            + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS,
 717                                                                                                    sep,
 718                                                                                                    ForesterUtil.LINE_SEPARATOR,
 719                                                                                                    BinaryDomainCombination.OutputFormat.DOT );
 720             }
 721             // HTML:
 722             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 723                                            go_id_to_term_map,
 724                                            go_namespace_limit,
 725                                            true,
 726                                            domain_parsimony.getGainLossMatrix(),
 727                                            CharacterStateMatrix.GainLossStates.GAIN,
 728                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC,
 729                                            sep,
 730                                            ForesterUtil.LINE_SEPARATOR,
 731                                            "Fitch Parsimony | Gains | Domain Combinations",
 732                                            "+",
 733                                            null,
 734                                            all_pfams_encountered,
 735                                            all_pfams_gained_as_dom_combinations,
 736                                            "_fitch_gains_dc" );
 737             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 738                                            go_id_to_term_map,
 739                                            go_namespace_limit,
 740                                            true,
 741                                            domain_parsimony.getGainLossMatrix(),
 742                                            CharacterStateMatrix.GainLossStates.LOSS,
 743                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC,
 744                                            sep,
 745                                            ForesterUtil.LINE_SEPARATOR,
 746                                            "Fitch Parsimony | Losses | Domain Combinations",
 747                                            "-",
 748                                            null,
 749                                            all_pfams_encountered,
 750                                            all_pfams_lost_as_dom_combinations,
 751                                            "_fitch_losses_dc" );
 752             writeBinaryStatesMatrixToList( domain_id_to_go_ids_map,
 753                                            go_id_to_term_map,
 754                                            go_namespace_limit,
 755                                            true,
 756                                            domain_parsimony.getGainLossMatrix(),
 757                                            null,
 758                                            outfile_name + surfacing.PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC,
 759                                            sep,
 760                                            ForesterUtil.LINE_SEPARATOR,
 761                                            "Fitch Parsimony | Present | Domain Combinations",
 762                                            "",
 763                                            null,
 764                                            all_pfams_encountered,
 765                                            null,
 766                                            "_fitch_present_dc" );
 767             writeAllEncounteredPfamsToFile( domain_id_to_go_ids_map,
 768                                             go_id_to_term_map,
 769                                             outfile_name,
 770                                             all_pfams_encountered );
 771             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX, all_pfams_gained_as_domains );
 772             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX, all_pfams_lost_as_domains );
 773             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_GAINED_AS_DC_SUFFIX,
 774                               all_pfams_gained_as_dom_combinations );
 775             writePfamsToFile( outfile_name + surfacing.ALL_PFAMS_LOST_AS_DC_SUFFIX, all_pfams_lost_as_dom_combinations );
 776             preparePhylogeny( local_phylogeny_l,
 777                               domain_parsimony,
 778                               date_time,
 779                               "Fitch parsimony on binary domain combination presence/absence randomization: "
 780                                       + randomization,
 781                               "fitch_on_binary_domain_combinations_" + outfile_name,
 782                               parameters_str );
 783             SurfacingUtil.writePhylogenyToFile( local_phylogeny_l, outfile_name
 784                     + surfacing.BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH );
 785             calculateIndependentDomainCombinationGains( local_phylogeny_l, outfile_name
 786                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX, outfile_name
 787                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX, outfile_name
 788                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX, outfile_name
 789                     + surfacing.INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX, outfile_name
 790                     + "_indep_dc_gains_fitch_lca_ranks.txt", outfile_name + "_indep_dc_gains_fitch_lca_taxonomies.txt" );
 791         }
 792     }
 793
 794     public static void executeParsimonyAnalysisForSecondaryFeatures( final String outfile_name,
 795                                                                      final DomainParsimonyCalculator secondary_features_parsimony,
 796                                                                      final Phylogeny phylogeny,
 797                                                                      final String parameters_str,
 798                                                                      final Map<Species, MappingResults> mapping_results_map ) {
 799         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 800         final String date_time = ForesterUtil.getCurrentDateTime();
 801         System.out.println();
 802         writeToNexus( outfile_name + surfacing.NEXUS_SECONDARY_FEATURES,
 803                       secondary_features_parsimony.createMatrixOfSecondaryFeaturePresenceOrAbsence( null ),
 804                       phylogeny );
 805         final Phylogeny local_phylogeny_copy = phylogeny.copy();
 806         secondary_features_parsimony.executeDolloParsimonyOnSecondaryFeatures( mapping_results_map );
 807         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossMatrix(), outfile_name
 808                 + surfacing.PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 809         SurfacingUtil.writeMatrixToFile( secondary_features_parsimony.getGainLossCountsMatrix(), outfile_name
 810                 + surfacing.PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES, Format.FORESTER );
 811         SurfacingUtil
 812                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 813                                                       CharacterStateMatrix.GainLossStates.GAIN,
 814                                                       outfile_name
 815                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES,
 816                                                       sep,
 817                                                       ForesterUtil.LINE_SEPARATOR,
 818                                                       null );
 819         SurfacingUtil
 820                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 821                                                       CharacterStateMatrix.GainLossStates.LOSS,
 822                                                       outfile_name
 823                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES,
 824                                                       sep,
 825                                                       ForesterUtil.LINE_SEPARATOR,
 826                                                       null );
 827         SurfacingUtil
 828                 .writeBinaryStatesMatrixAsListToFile( secondary_features_parsimony.getGainLossMatrix(),
 829                                                       null,
 830                                                       outfile_name
 831                                                               + surfacing.PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES,
 832                                                       sep,
 833                                                       ForesterUtil.LINE_SEPARATOR,
 834                                                       null );
 835         preparePhylogeny( local_phylogeny_copy,
 836                           secondary_features_parsimony,
 837                           date_time,
 838                           "Dollo parsimony on secondary feature presence/absence",
 839                           "dollo_on_secondary_features_" + outfile_name,
 840                           parameters_str );
 841         SurfacingUtil.writePhylogenyToFile( local_phylogeny_copy, outfile_name
 842                 + surfacing.SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO );
 843     }
 844
 845     public static void extractProteinNames( final List<Protein> proteins,
 846                                             final List<DomainId> query_domain_ids_nc_order,
 847                                             final Writer out,
 848                                             final String separator,
 849                                             final String limit_to_species ) throws IOException {
 850         for( final Protein protein : proteins ) {
 851             if ( ForesterUtil.isEmpty( limit_to_species )
 852                     || protein.getSpecies().getSpeciesId().equalsIgnoreCase( limit_to_species ) ) {
 853                 if ( protein.contains( query_domain_ids_nc_order, true ) ) {
 854                     out.write( protein.getSpecies().getSpeciesId() );
 855                     out.write( separator );
 856                     out.write( protein.getProteinId().getId() );
 857                     out.write( separator );
 858                     out.write( "[" );
 859                     final Set<DomainId> visited_domain_ids = new HashSet<DomainId>();
 860                     boolean first = true;
 861                     for( final Domain domain : protein.getProteinDomains() ) {
 862                         if ( !visited_domain_ids.contains( domain.getDomainId() ) ) {
 863                             visited_domain_ids.add( domain.getDomainId() );
 864                             if ( first ) {
 865                                 first = false;
 866                             }
 867                             else {
 868                                 out.write( " " );
 869                             }
 870                             out.write( domain.getDomainId().getId() );
 871                             out.write( " {" );
 872                             out.write( "" + domain.getTotalCount() );
 873                             out.write( "}" );
 874                         }
 875                     }
 876                     out.write( "]" );
 877                     out.write( separator );
 878                     if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 879                             .equals( SurfacingConstants.NONE ) ) ) {
 880                         out.write( protein.getDescription() );
 881                     }
 882                     out.write( separator );
 883                     if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 884                             .equals( SurfacingConstants.NONE ) ) ) {
 885                         out.write( protein.getAccession() );
 886                     }
 887                     out.write( SurfacingConstants.NL );
 888                 }
 889             }
 890         }
 891         out.flush();
 892     }
 893
 894     public static void extractProteinNames( final SortedMap<Species, List<Protein>> protein_lists_per_species,
 895                                             final DomainId domain_id,
 896                                             final Writer out,
 897                                             final String separator,
 898                                             final String limit_to_species ) throws IOException {
 899         for( final Species species : protein_lists_per_species.keySet() ) {
 900             for( final Protein protein : protein_lists_per_species.get( species ) ) {
 901                 if ( ForesterUtil.isEmpty( limit_to_species )
 902                         || protein.getSpecies().getSpeciesId().equalsIgnoreCase( limit_to_species ) ) {
 903                     final List<Domain> domains = protein.getProteinDomains( domain_id );
 904                     if ( domains.size() > 0 ) {
 905                         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 906                         for( final Domain domain : domains ) {
 907                             stats.addValue( domain.getPerSequenceEvalue() );
 908                         }
 909                         out.write( protein.getSpecies().getSpeciesId() );
 910                         out.write( separator );
 911                         out.write( protein.getProteinId().getId() );
 912                         out.write( separator );
 913                         out.write( "[" + FORMATTER.format( stats.median() ) + "]" );
 914                         out.write( separator );
 915                         if ( !( ForesterUtil.isEmpty( protein.getDescription() ) || protein.getDescription()
 916                                 .equals( SurfacingConstants.NONE ) ) ) {
 917                             out.write( protein.getDescription() );
 918                         }
 919                         out.write( separator );
 920                         if ( !( ForesterUtil.isEmpty( protein.getAccession() ) || protein.getAccession()
 921                                 .equals( SurfacingConstants.NONE ) ) ) {
 922                             out.write( protein.getAccession() );
 923                         }
 924                         out.write( SurfacingConstants.NL );
 925                     }
 926                 }
 927             }
 928         }
 929         out.flush();
 930     }
 931
 932     public static SortedSet<DomainId> getAllDomainIds( final List<GenomeWideCombinableDomains> gwcd_list ) {
 933         final SortedSet<DomainId> all_domains_ids = new TreeSet<DomainId>();
 934         for( final GenomeWideCombinableDomains gwcd : gwcd_list ) {
 935             final Set<DomainId> all_domains = gwcd.getAllDomainIds();
 936             //    for( final Domain domain : all_domains ) {
 937             all_domains_ids.addAll( all_domains );
 938             //    }
 939         }
 940         return all_domains_ids;
 941     }
 942
 943     public static SortedMap<String, Integer> getDomainCounts( final List<Protein> protein_domain_collections ) {
 944         final SortedMap<String, Integer> map = new TreeMap<String, Integer>();
 945         for( final Protein protein_domain_collection : protein_domain_collections ) {
 946             for( final Object name : protein_domain_collection.getProteinDomains() ) {
 947                 final BasicDomain protein_domain = ( BasicDomain ) name;
 948                 final String id = protein_domain.getDomainId().getId();
 949                 if ( map.containsKey( id ) ) {
 950                     map.put( id, map.get( id ) + 1 );
 951                 }
 952                 else {
 953                     map.put( id, 1 );
 954                 }
 955             }
 956         }
 957         return map;
 958     }
 959
 960     public static int getNumberOfNodesLackingName( final Phylogeny p, final StringBuilder names ) {
 961         final PhylogenyNodeIterator it = p.iteratorPostorder();
 962         int c = 0;
 963         while ( it.hasNext() ) {
 964             final PhylogenyNode n = it.next();
 965             if ( ForesterUtil.isEmpty( n.getName() )
 966                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
 967                             .getScientificName() ) )
 968                     && ( !n.getNodeData().isHasTaxonomy() || ForesterUtil.isEmpty( n.getNodeData().getTaxonomy()
 969                             .getCommonName() ) ) ) {
 970                 if ( n.getParent() != null ) {
 971                     names.append( " " );
 972                     names.append( n.getParent().getName() );
 973                 }
 974                 final List l = n.getAllExternalDescendants();
 975                 for( final Object object : l ) {
 976                     System.out.println( l.toString() );
 977                 }
 978                 ++c;
 979             }
 980         }
 981         return c;
 982     }
 983
 984     /**
 985      * Returns true is Domain domain falls in an uninterrupted stretch of
 986      * covered positions.
 987      *
 988      * @param domain
 989      * @param covered_positions
 990      * @return
 991      */
 992     public static boolean isEngulfed( final Domain domain, final List<Boolean> covered_positions ) {
 993         for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
 994             if ( ( i >= covered_positions.size() ) || ( covered_positions.get( i ) != true ) ) {
 995                 return false;
 996             }
 997         }
 998         return true;
 999     }
1000
1001     public static void preparePhylogeny( final Phylogeny p,
1002                                          final DomainParsimonyCalculator domain_parsimony,
1003                                          final String date_time,
1004                                          final String method,
1005                                          final String name,
1006                                          final String parameters_str ) {
1007         domain_parsimony.decoratePhylogenyWithDomains( p );
1008         final StringBuilder desc = new StringBuilder();
1009         desc.append( "[Method: " + method + "] [Date: " + date_time + "] " );
1010         desc.append( "[Cost: " + domain_parsimony.getCost() + "] " );
1011         desc.append( "[Gains: " + domain_parsimony.getTotalGains() + "] " );
1012         desc.append( "[Losses: " + domain_parsimony.getTotalLosses() + "] " );
1013         desc.append( "[Unchanged: " + domain_parsimony.getTotalUnchanged() + "] " );
1014         desc.append( "[Parameters: " + parameters_str + "]" );
1015         p.setName( name );
1016         p.setDescription( desc.toString() );
1017         p.setConfidence( new Confidence( domain_parsimony.getCost(), "parsimony" ) );
1018         p.setRerootable( false );
1019         p.setRooted( true );
1020     }
1021
1022     /*
1023      * species | protein id | n-terminal domain | c-terminal domain | n-terminal domain per domain E-value | c-terminal domain per domain E-value
1024      *
1025      *
1026      */
1027     static public StringBuffer proteinToDomainCombinations( final Protein protein,
1028                                                             final String protein_id,
1029                                                             final String separator ) {
1030         final StringBuffer sb = new StringBuffer();
1031         if ( protein.getSpecies() == null ) {
1032             throw new IllegalArgumentException( "species must not be null" );
1033         }
1034         if ( ForesterUtil.isEmpty( protein.getSpecies().getSpeciesId() ) ) {
1035             throw new IllegalArgumentException( "species id must not be empty" );
1036         }
1037         final List<Domain> domains = protein.getProteinDomains();
1038         if ( domains.size() > 1 ) {
1039             final Map<String, Integer> counts = new HashMap<String, Integer>();
1040             for( final Domain domain : domains ) {
1041                 final String id = domain.getDomainId().getId();
1042                 if ( counts.containsKey( id ) ) {
1043                     counts.put( id, counts.get( id ) + 1 );
1044                 }
1045                 else {
1046                     counts.put( id, 1 );
1047                 }
1048             }
1049             final Set<String> dcs = new HashSet<String>();
1050             for( int i = 1; i < domains.size(); ++i ) {
1051                 for( int j = 0; j < i; ++j ) {
1052                     Domain domain_n = domains.get( i );
1053                     Domain domain_c = domains.get( j );
1054                     if ( domain_n.getFrom() > domain_c.getFrom() ) {
1055                         domain_n = domains.get( j );
1056                         domain_c = domains.get( i );
1057                     }
1058                     final String dc = domain_n.getDomainId().getId() + domain_c.getDomainId().getId();
1059                     if ( !dcs.contains( dc ) ) {
1060                         dcs.add( dc );
1061                         sb.append( protein.getSpecies() );
1062                         sb.append( separator );
1063                         sb.append( protein_id );
1064                         sb.append( separator );
1065                         sb.append( domain_n.getDomainId().getId() );
1066                         sb.append( separator );
1067                         sb.append( domain_c.getDomainId().getId() );
1068                         sb.append( separator );
1069                         sb.append( domain_n.getPerDomainEvalue() );
1070                         sb.append( separator );
1071                         sb.append( domain_c.getPerDomainEvalue() );
1072                         sb.append( separator );
1073                         sb.append( counts.get( domain_n.getDomainId().getId() ) );
1074                         sb.append( separator );
1075                         sb.append( counts.get( domain_c.getDomainId().getId() ) );
1076                         sb.append( ForesterUtil.LINE_SEPARATOR );
1077                     }
1078                 }
1079             }
1080         }
1081         else if ( domains.size() == 1 ) {
1082             sb.append( protein.getSpecies() );
1083             sb.append( separator );
1084             sb.append( protein_id );
1085             sb.append( separator );
1086             sb.append( domains.get( 0 ).getDomainId().getId() );
1087             sb.append( separator );
1088             sb.append( separator );
1089             sb.append( domains.get( 0 ).getPerDomainEvalue() );
1090             sb.append( separator );
1091             sb.append( separator );
1092             sb.append( 1 );
1093             sb.append( separator );
1094             sb.append( ForesterUtil.LINE_SEPARATOR );
1095         }
1096         else {
1097             sb.append( protein.getSpecies() );
1098             sb.append( separator );
1099             sb.append( protein_id );
1100             sb.append( separator );
1101             sb.append( separator );
1102             sb.append( separator );
1103             sb.append( separator );
1104             sb.append( separator );
1105             sb.append( separator );
1106             sb.append( ForesterUtil.LINE_SEPARATOR );
1107         }
1108         return sb;
1109     }
1110
1111     /**
1112      *
1113      * Example regarding engulfment: ------------0.1 ----------0.2 --0.3 =>
1114      * domain with 0.3 is ignored
1115      *
1116      * -----------0.1 ----------0.2 --0.3 => domain with 0.3 is ignored
1117      *
1118      *
1119      * ------------0.1 ----------0.3 --0.2 => domains with 0.3 and 0.2 are _not_
1120      * ignored
1121      *
1122      * @param max_allowed_overlap
1123      *            maximal allowed overlap (inclusive) to be still considered not
1124      *            overlapping (zero or negative value to allow any overlap)
1125      * @param remove_engulfed_domains
1126      *            to remove domains which are completely engulfed by coverage of
1127      *            domains with better support
1128      * @param protein
1129      * @return
1130      */
1131     public static Protein removeOverlappingDomains( final int max_allowed_overlap,
1132                                                     final boolean remove_engulfed_domains,
1133                                                     final Protein protein ) {
1134         final Protein pruned_protein = new BasicProtein( protein.getProteinId().getId(), protein.getSpecies()
1135                 .getSpeciesId() );
1136         final List<Domain> sorted = SurfacingUtil.sortDomainsWithAscendingConfidenceValues( protein );
1137         final List<Boolean> covered_positions = new ArrayList<Boolean>();
1138         for( final Domain domain : sorted ) {
1139             if ( ( ( max_allowed_overlap < 0 ) || ( SurfacingUtil.calculateOverlap( domain, covered_positions ) <= max_allowed_overlap ) )
1140                     && ( !remove_engulfed_domains || !isEngulfed( domain, covered_positions ) ) ) {
1141                 final int covered_positions_size = covered_positions.size();
1142                 for( int i = covered_positions_size; i < domain.getFrom(); ++i ) {
1143                     covered_positions.add( false );
1144                 }
1145                 final int new_covered_positions_size = covered_positions.size();
1146                 for( int i = domain.getFrom(); i <= domain.getTo(); ++i ) {
1147                     if ( i < new_covered_positions_size ) {
1148                         covered_positions.set( i, true );
1149                     }
1150                     else {
1151                         covered_positions.add( true );
1152                     }
1153                 }
1154                 pruned_protein.addProteinDomain( domain );
1155             }
1156         }
1157         return pruned_protein;
1158     }
1159
1160     public static List<Domain> sortDomainsWithAscendingConfidenceValues( final Protein protein ) {
1161         final List<Domain> domains = new ArrayList<Domain>();
1162         for( final Domain d : protein.getProteinDomains() ) {
1163             domains.add( d );
1164         }
1165         Collections.sort( domains, SurfacingUtil.ASCENDING_CONFIDENCE_VALUE_ORDER );
1166         return domains;
1167     }
1168
1169     private static List<String> splitDomainCombination( final String dc ) {
1170         final String[] s = dc.split( "=" );
1171         if ( s.length != 2 ) {
1172             ForesterUtil.printErrorMessage( surfacing.PRG_NAME, "Stringyfied domain combination has illegal format: "
1173                     + dc );
1174             System.exit( -1 );
1175         }
1176         final List<String> l = new ArrayList<String>( 2 );
1177         l.add( s[ 0 ] );
1178         l.add( s[ 1 ] );
1179         return l;
1180     }
1181
1182     public static void writeAllDomainsChangedOnAllSubtrees( final Phylogeny p,
1183                                                             final boolean get_gains,
1184                                                             final String outdir,
1185                                                             final String suffix_for_filename ) throws IOException {
1186         CharacterStateMatrix.GainLossStates state = CharacterStateMatrix.GainLossStates.GAIN;
1187         if ( !get_gains ) {
1188             state = CharacterStateMatrix.GainLossStates.LOSS;
1189         }
1190         final File base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES,
1191                                                                   false,
1192                                                                   state,
1193                                                                   outdir );
1194         for( final PhylogenyNodeIterator it = p.iteratorPostorder(); it.hasNext(); ) {
1195             final PhylogenyNode node = it.next();
1196             if ( !node.isExternal() ) {
1197                 final SortedSet<String> domains = collectAllDomainsChangedOnSubtree( node, get_gains );
1198                 if ( domains.size() > 0 ) {
1199                     final Writer writer = ForesterUtil.createBufferedWriter( base_dir + ForesterUtil.FILE_SEPARATOR
1200                             + node.getName() + suffix_for_filename );
1201                     for( final String domain : domains ) {
1202                         writer.write( domain );
1203                         writer.write( ForesterUtil.LINE_SEPARATOR );
1204                     }
1205                     writer.close();
1206                 }
1207             }
1208         }
1209     }
1210
1211     private static void writeAllEncounteredPfamsToFile( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1212                                                         final Map<GoId, GoTerm> go_id_to_term_map,
1213                                                         final String outfile_name,
1214                                                         final SortedSet<String> all_pfams_encountered ) {
1215         final File all_pfams_encountered_file = new File( outfile_name + surfacing.ALL_PFAMS_ENCOUNTERED_SUFFIX );
1216         final File all_pfams_encountered_with_go_annotation_file = new File( outfile_name
1217                 + surfacing.ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX );
1218         final File encountered_pfams_summary_file = new File( outfile_name + surfacing.ENCOUNTERED_PFAMS_SUMMARY_SUFFIX );
1219         int biological_process_counter = 0;
1220         int cellular_component_counter = 0;
1221         int molecular_function_counter = 0;
1222         int pfams_with_mappings_counter = 0;
1223         int pfams_without_mappings_counter = 0;
1224         int pfams_without_mappings_to_bp_or_mf_counter = 0;
1225         int pfams_with_mappings_to_bp_or_mf_counter = 0;
1226         try {
1227             final Writer all_pfams_encountered_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_file ) );
1228             final Writer all_pfams_encountered_with_go_annotation_writer = new BufferedWriter( new FileWriter( all_pfams_encountered_with_go_annotation_file ) );
1229             final Writer summary_writer = new BufferedWriter( new FileWriter( encountered_pfams_summary_file ) );
1230             summary_writer.write( "# Pfam to GO mapping summary" );
1231             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1232             summary_writer.write( "# Actual summary is at the end of this file." );
1233             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1234             summary_writer.write( "# Encountered Pfams without a GO mapping:" );
1235             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1236             for( final String pfam : all_pfams_encountered ) {
1237                 all_pfams_encountered_writer.write( pfam );
1238                 all_pfams_encountered_writer.write( ForesterUtil.LINE_SEPARATOR );
1239                 final DomainId domain_id = new DomainId( pfam );
1240                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1241                     ++pfams_with_mappings_counter;
1242                     all_pfams_encountered_with_go_annotation_writer.write( pfam );
1243                     all_pfams_encountered_with_go_annotation_writer.write( ForesterUtil.LINE_SEPARATOR );
1244                     final List<GoId> go_ids = domain_id_to_go_ids_map.get( domain_id );
1245                     boolean maps_to_bp = false;
1246                     boolean maps_to_cc = false;
1247                     boolean maps_to_mf = false;
1248                     for( final GoId go_id : go_ids ) {
1249                         final GoTerm go_term = go_id_to_term_map.get( go_id );
1250                         if ( go_term.getGoNameSpace().isBiologicalProcess() ) {
1251                             maps_to_bp = true;
1252                         }
1253                         else if ( go_term.getGoNameSpace().isCellularComponent() ) {
1254                             maps_to_cc = true;
1255                         }
1256                         else if ( go_term.getGoNameSpace().isMolecularFunction() ) {
1257                             maps_to_mf = true;
1258                         }
1259                     }
1260                     if ( maps_to_bp ) {
1261                         ++biological_process_counter;
1262                     }
1263                     if ( maps_to_cc ) {
1264                         ++cellular_component_counter;
1265                     }
1266                     if ( maps_to_mf ) {
1267                         ++molecular_function_counter;
1268                     }
1269                     if ( maps_to_bp || maps_to_mf ) {
1270                         ++pfams_with_mappings_to_bp_or_mf_counter;
1271                     }
1272                     else {
1273                         ++pfams_without_mappings_to_bp_or_mf_counter;
1274                     }
1275                 }
1276                 else {
1277                     ++pfams_without_mappings_to_bp_or_mf_counter;
1278                     ++pfams_without_mappings_counter;
1279                     summary_writer.write( pfam );
1280                     summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1281                 }
1282             }
1283             all_pfams_encountered_writer.close();
1284             all_pfams_encountered_with_go_annotation_writer.close();
1285             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + all_pfams_encountered.size()
1286                     + "] encountered Pfams to: \"" + all_pfams_encountered_file + "\"" );
1287             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote all [" + pfams_with_mappings_counter
1288                     + "] encountered Pfams with GO mappings to: \"" + all_pfams_encountered_with_go_annotation_file
1289                     + "\"" );
1290             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote summary (including all ["
1291                     + pfams_without_mappings_counter + "] encountered Pfams without GO mappings) to: \""
1292                     + encountered_pfams_summary_file + "\"" );
1293             ForesterUtil.programMessage( surfacing.PRG_NAME, "Sum of Pfams encountered                : "
1294                     + all_pfams_encountered.size() );
1295             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without a mapping                 : "
1296                     + pfams_without_mappings_counter + " ["
1297                     + ( 100 * pfams_without_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1298             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams without mapping to proc. or func. : "
1299                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
1300                     + ( 100 * pfams_without_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1301             ForesterUtil.programMessage( surfacing.PRG_NAME,
1302                                          "Pfams with a mapping                    : " + pfams_with_mappings_counter
1303                                                  + " ["
1304                                                  + ( 100 * pfams_with_mappings_counter / all_pfams_encountered.size() )
1305                                                  + "%]" );
1306             ForesterUtil.programMessage( surfacing.PRG_NAME, "Pfams with a mapping to proc. or func.  : "
1307                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
1308                     + ( 100 * pfams_with_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1309             ForesterUtil.programMessage( surfacing.PRG_NAME,
1310                                          "Pfams with mapping to biological process: " + biological_process_counter
1311                                                  + " ["
1312                                                  + ( 100 * biological_process_counter / all_pfams_encountered.size() )
1313                                                  + "%]" );
1314             ForesterUtil.programMessage( surfacing.PRG_NAME,
1315                                          "Pfams with mapping to molecular function: " + molecular_function_counter
1316                                                  + " ["
1317                                                  + ( 100 * molecular_function_counter / all_pfams_encountered.size() )
1318                                                  + "%]" );
1319             ForesterUtil.programMessage( surfacing.PRG_NAME,
1320                                          "Pfams with mapping to cellular component: " + cellular_component_counter
1321                                                  + " ["
1322                                                  + ( 100 * cellular_component_counter / all_pfams_encountered.size() )
1323                                                  + "%]" );
1324             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1325             summary_writer.write( "# Sum of Pfams encountered                : " + all_pfams_encountered.size() );
1326             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1327             summary_writer.write( "# Pfams without a mapping                 : " + pfams_without_mappings_counter
1328                     + " [" + ( 100 * pfams_without_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1329             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1330             summary_writer.write( "# Pfams without mapping to proc. or func. : "
1331                     + pfams_without_mappings_to_bp_or_mf_counter + " ["
1332                     + ( 100 * pfams_without_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1333             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1334             summary_writer.write( "# Pfams with a mapping                    : " + pfams_with_mappings_counter + " ["
1335                     + ( 100 * pfams_with_mappings_counter / all_pfams_encountered.size() ) + "%]" );
1336             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1337             summary_writer.write( "# Pfams with a mapping to proc. or func.  : "
1338                     + pfams_with_mappings_to_bp_or_mf_counter + " ["
1339                     + ( 100 * pfams_with_mappings_to_bp_or_mf_counter / all_pfams_encountered.size() ) + "%]" );
1340             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1341             summary_writer.write( "# Pfams with mapping to biological process: " + biological_process_counter + " ["
1342                     + ( 100 * biological_process_counter / all_pfams_encountered.size() ) + "%]" );
1343             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1344             summary_writer.write( "# Pfams with mapping to molecular function: " + molecular_function_counter + " ["
1345                     + ( 100 * molecular_function_counter / all_pfams_encountered.size() ) + "%]" );
1346             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1347             summary_writer.write( "# Pfams with mapping to cellular component: " + cellular_component_counter + " ["
1348                     + ( 100 * cellular_component_counter / all_pfams_encountered.size() ) + "%]" );
1349             summary_writer.write( ForesterUtil.LINE_SEPARATOR );
1350             summary_writer.close();
1351         }
1352         catch ( final IOException e ) {
1353             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
1354         }
1355     }
1356
1357     public static void writeBinaryDomainCombinationsFileForGraphAnalysis( final String[][] input_file_properties,
1358                                                                           final File output_dir,
1359                                                                           final GenomeWideCombinableDomains gwcd,
1360                                                                           final int i,
1361                                                                           final GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1362         File dc_outfile_dot = new File( input_file_properties[ i ][ 0 ]
1363                 + surfacing.DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS );
1364         if ( output_dir != null ) {
1365             dc_outfile_dot = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile_dot );
1366         }
1367         checkForOutputFileWriteability( dc_outfile_dot );
1368         final SortedSet<BinaryDomainCombination> binary_combinations = createSetOfAllBinaryDomainCombinationsPerGenome( gwcd );
1369         try {
1370             final BufferedWriter out_dot = new BufferedWriter( new FileWriter( dc_outfile_dot ) );
1371             for( final BinaryDomainCombination bdc : binary_combinations ) {
1372                 out_dot.write( bdc.toGraphDescribingLanguage( BinaryDomainCombination.OutputFormat.DOT, null, null )
1373                         .toString() );
1374                 out_dot.write( SurfacingConstants.NL );
1375             }
1376             out_dot.close();
1377         }
1378         catch ( final IOException e ) {
1379             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1380         }
1381         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote binary domain combination for \""
1382                 + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1383                 + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile_dot + "\"" );
1384     }
1385
1386     public static void writeBinaryStatesMatrixAsListToFile( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1387                                                             final CharacterStateMatrix.GainLossStates state,
1388                                                             final String filename,
1389                                                             final String indentifier_characters_separator,
1390                                                             final String character_separator,
1391                                                             final Map<String, String> descriptions ) {
1392         final File outfile = new File( filename );
1393         checkForOutputFileWriteability( outfile );
1394         final SortedSet<String> sorted_ids = new TreeSet<String>();
1395         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1396             sorted_ids.add( matrix.getIdentifier( i ) );
1397         }
1398         try {
1399             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1400             for( final String id : sorted_ids ) {
1401                 out.write( indentifier_characters_separator );
1402                 out.write( "#" + id );
1403                 out.write( indentifier_characters_separator );
1404                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1405                     // Not nice:
1406                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1407                     if ( ( matrix.getState( id, c ) == state )
1408                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1409                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1410                         out.write( matrix.getCharacter( c ) );
1411                         if ( ( descriptions != null ) && !descriptions.isEmpty()
1412                                 && descriptions.containsKey( matrix.getCharacter( c ) ) ) {
1413                             out.write( "\t" );
1414                             out.write( descriptions.get( matrix.getCharacter( c ) ) );
1415                         }
1416                         out.write( character_separator );
1417                     }
1418                 }
1419             }
1420             out.flush();
1421             out.close();
1422         }
1423         catch ( final IOException e ) {
1424             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1425         }
1426         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1427     }
1428
1429     public static void writeBinaryStatesMatrixAsListToFileForBinaryCombinationsForGraphAnalysis( final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1430                                                                                                  final CharacterStateMatrix.GainLossStates state,
1431                                                                                                  final String filename,
1432                                                                                                  final String indentifier_characters_separator,
1433                                                                                                  final String character_separator,
1434                                                                                                  final BinaryDomainCombination.OutputFormat bc_output_format ) {
1435         final File outfile = new File( filename );
1436         checkForOutputFileWriteability( outfile );
1437         final SortedSet<String> sorted_ids = new TreeSet<String>();
1438         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1439             sorted_ids.add( matrix.getIdentifier( i ) );
1440         }
1441         try {
1442             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
1443             for( final String id : sorted_ids ) {
1444                 out.write( indentifier_characters_separator );
1445                 out.write( "#" + id );
1446                 out.write( indentifier_characters_separator );
1447                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1448                     // Not nice:
1449                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1450                     if ( ( matrix.getState( id, c ) == state )
1451                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) || ( matrix
1452                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) ) ) ) {
1453                         BinaryDomainCombination bdc = null;
1454                         try {
1455                             bdc = BasicBinaryDomainCombination.createInstance( matrix.getCharacter( c ) );
1456                         }
1457                         catch ( final Exception e ) {
1458                             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getLocalizedMessage() );
1459                         }
1460                         out.write( bdc.toGraphDescribingLanguage( bc_output_format, null, null ).toString() );
1461                         out.write( character_separator );
1462                     }
1463                 }
1464             }
1465             out.flush();
1466             out.close();
1467         }
1468         catch ( final IOException e ) {
1469             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1470         }
1471         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters list: \"" + filename + "\"" );
1472     }
1473
1474     public static void writeBinaryStatesMatrixToList( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1475                                                       final Map<GoId, GoTerm> go_id_to_term_map,
1476                                                       final GoNameSpace go_namespace_limit,
1477                                                       final boolean domain_combinations,
1478                                                       final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1479                                                       final CharacterStateMatrix.GainLossStates state,
1480                                                       final String filename,
1481                                                       final String indentifier_characters_separator,
1482                                                       final String character_separator,
1483                                                       final String title_for_html,
1484                                                       final String prefix_for_html,
1485                                                       final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1486                                                       final SortedSet<String> all_pfams_encountered,
1487                                                       final SortedSet<String> pfams_gained_or_lost,
1488                                                       final String suffix_for_per_node_events_file ) {
1489         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1490             throw new IllegalArgumentException( "attempt to use GO namespace limit without a GO-id to term map" );
1491         }
1492         else if ( ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) ) {
1493             throw new IllegalArgumentException( "attempt to output detailed HTML without a Pfam to GO map" );
1494         }
1495         else if ( ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1496             throw new IllegalArgumentException( "attempt to output detailed HTML without a GO-id to term map" );
1497         }
1498         final File outfile = new File( filename );
1499         checkForOutputFileWriteability( outfile );
1500         final SortedSet<String> sorted_ids = new TreeSet<String>();
1501         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1502             sorted_ids.add( matrix.getIdentifier( i ) );
1503         }
1504         try {
1505             final Writer out = new BufferedWriter( new FileWriter( outfile ) );
1506             final File per_node_go_mapped_domain_gain_loss_files_base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES,
1507                                                                                                                 domain_combinations,
1508                                                                                                                 state,
1509                                                                                                                 filename );
1510             Writer per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1511             File per_node_go_mapped_domain_gain_loss_outfile = null;
1512             int per_node_counter = 0;
1513             out.write( "<html>" );
1514             out.write( SurfacingConstants.NL );
1515             addHtmlHead( out, title_for_html );
1516             out.write( SurfacingConstants.NL );
1517             out.write( "<body>" );
1518             out.write( SurfacingConstants.NL );
1519             out.write( "<h1>" );
1520             out.write( SurfacingConstants.NL );
1521             out.write( title_for_html );
1522             out.write( SurfacingConstants.NL );
1523             out.write( "</h1>" );
1524             out.write( SurfacingConstants.NL );
1525             out.write( "<table>" );
1526             out.write( SurfacingConstants.NL );
1527             for( final String id : sorted_ids ) {
1528                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1529                 if ( matcher.matches() ) {
1530                     continue;
1531                 }
1532                 out.write( "<tr>" );
1533                 out.write( "<td>" );
1534                 out.write( "<a href=\"#" + id + "\">" + id + "</a>" );
1535                 out.write( "</td>" );
1536                 out.write( "</tr>" );
1537                 out.write( SurfacingConstants.NL );
1538             }
1539             out.write( "</table>" );
1540             out.write( SurfacingConstants.NL );
1541             for( final String id : sorted_ids ) {
1542                 final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( id );
1543                 if ( matcher.matches() ) {
1544                     continue;
1545                 }
1546                 out.write( SurfacingConstants.NL );
1547                 out.write( "<h2>" );
1548                 out.write( "<a name=\"" + id + "\">" + id + "</a>" );
1549                 writeTaxonomyLinks( out, id );
1550                 out.write( "</h2>" );
1551                 out.write( SurfacingConstants.NL );
1552                 out.write( "<table>" );
1553                 out.write( SurfacingConstants.NL );
1554                 out.write( "<tr>" );
1555                 out.write( "<td><b>" );
1556                 out.write( "Pfam domain(s)" );
1557                 out.write( "</b></td><td><b>" );
1558                 out.write( "GO term acc" );
1559                 out.write( "</b></td><td><b>" );
1560                 out.write( "GO term" );
1561                 out.write( "</b></td><td><b>" );
1562                 out.write( "GO namespace" );
1563                 out.write( "</b></td>" );
1564                 out.write( "</tr>" );
1565                 out.write( SurfacingConstants.NL );
1566                 out.write( "</tr>" );
1567                 out.write( SurfacingConstants.NL );
1568                 per_node_counter = 0;
1569                 if ( matrix.getNumberOfCharacters() > 0 ) {
1570                     per_node_go_mapped_domain_gain_loss_outfile = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
1571                             + ForesterUtil.FILE_SEPARATOR + id + suffix_for_per_node_events_file );
1572                     SurfacingUtil.checkForOutputFileWriteability( per_node_go_mapped_domain_gain_loss_outfile );
1573                     per_node_go_mapped_domain_gain_loss_outfile_writer = ForesterUtil
1574                             .createBufferedWriter( per_node_go_mapped_domain_gain_loss_outfile );
1575                 }
1576                 else {
1577                     per_node_go_mapped_domain_gain_loss_outfile = null;
1578                     per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1579                 }
1580                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1581                     // Not nice:
1582                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1583                     if ( ( matrix.getState( id, c ) == state )
1584                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) || ( matrix
1585                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) ) ) ) {
1586                         final String character = matrix.getCharacter( c );
1587                         String domain_0 = "";
1588                         String domain_1 = "";
1589                         if ( character.indexOf( BinaryDomainCombination.SEPARATOR ) > 0 ) {
1590                             final String[] s = character.split( BinaryDomainCombination.SEPARATOR );
1591                             if ( s.length != 2 ) {
1592                                 throw new AssertionError( "this should not have happened: unexpected format for domain combination: ["
1593                                         + character + "]" );
1594                             }
1595                             domain_0 = s[ 0 ];
1596                             domain_1 = s[ 1 ];
1597                         }
1598                         else {
1599                             domain_0 = character;
1600                         }
1601                         writeDomainData( domain_id_to_go_ids_map,
1602                                          go_id_to_term_map,
1603                                          go_namespace_limit,
1604                                          out,
1605                                          domain_0,
1606                                          domain_1,
1607                                          prefix_for_html,
1608                                          character_separator,
1609                                          domain_id_to_secondary_features_maps,
1610                                          null );
1611                         all_pfams_encountered.add( domain_0 );
1612                         if ( pfams_gained_or_lost != null ) {
1613                             pfams_gained_or_lost.add( domain_0 );
1614                         }
1615                         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
1616                             all_pfams_encountered.add( domain_1 );
1617                             if ( pfams_gained_or_lost != null ) {
1618                                 pfams_gained_or_lost.add( domain_1 );
1619                             }
1620                         }
1621                         if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1622                             writeDomainsToIndividualFilePerTreeNode( per_node_go_mapped_domain_gain_loss_outfile_writer,
1623                                                                      domain_0,
1624                                                                      domain_1 );
1625                             per_node_counter++;
1626                         }
1627                     }
1628                 }
1629                 if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1630                     per_node_go_mapped_domain_gain_loss_outfile_writer.close();
1631                     if ( per_node_counter < 1 ) {
1632                         per_node_go_mapped_domain_gain_loss_outfile.delete();
1633                     }
1634                     per_node_counter = 0;
1635                 }
1636                 out.write( "</table>" );
1637                 out.write( SurfacingConstants.NL );
1638                 out.write( "<hr>" );
1639                 out.write( SurfacingConstants.NL );
1640             } // for( final String id : sorted_ids ) {
1641             out.write( "</body>" );
1642             out.write( SurfacingConstants.NL );
1643             out.write( "</html>" );
1644             out.write( SurfacingConstants.NL );
1645             out.flush();
1646             out.close();
1647         }
1648         catch ( final IOException e ) {
1649             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1650         }
1651         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters detailed HTML list: \"" + filename + "\"" );
1652     }
1653
1654     public static void writeBinaryStatesMatrixToListORIGIG( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1655                                                             final Map<GoId, GoTerm> go_id_to_term_map,
1656                                                             final GoNameSpace go_namespace_limit,
1657                                                             final boolean domain_combinations,
1658                                                             final CharacterStateMatrix<CharacterStateMatrix.GainLossStates> matrix,
1659                                                             final CharacterStateMatrix.GainLossStates state,
1660                                                             final String filename,
1661                                                             final String indentifier_characters_separator,
1662                                                             final String character_separator,
1663                                                             final String title_for_html,
1664                                                             final String prefix_for_html,
1665                                                             final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1666                                                             final SortedSet<String> all_pfams_encountered,
1667                                                             final SortedSet<String> pfams_gained_or_lost,
1668                                                             final String suffix_for_per_node_events_file ) {
1669         if ( ( go_namespace_limit != null ) && ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1670             throw new IllegalArgumentException( "attempt to use GO namespace limit without a GO-id to term map" );
1671         }
1672         else if ( ( ( domain_id_to_go_ids_map == null ) || ( domain_id_to_go_ids_map.size() < 1 ) ) ) {
1673             throw new IllegalArgumentException( "attempt to output detailed HTML without a Pfam to GO map" );
1674         }
1675         else if ( ( ( go_id_to_term_map == null ) || ( go_id_to_term_map.size() < 1 ) ) ) {
1676             throw new IllegalArgumentException( "attempt to output detailed HTML without a GO-id to term map" );
1677         }
1678         final File outfile = new File( filename );
1679         checkForOutputFileWriteability( outfile );
1680         final SortedSet<String> sorted_ids = new TreeSet<String>();
1681         for( int i = 0; i < matrix.getNumberOfIdentifiers(); ++i ) {
1682             sorted_ids.add( matrix.getIdentifier( i ) );
1683         }
1684         try {
1685             final Writer out = new BufferedWriter( new FileWriter( outfile ) );
1686             final File per_node_go_mapped_domain_gain_loss_files_base_dir = createBaseDirForPerNodeDomainFiles( surfacing.BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES,
1687                                                                                                                 domain_combinations,
1688                                                                                                                 state,
1689                                                                                                                 filename );
1690             Writer per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1691             File per_node_go_mapped_domain_gain_loss_outfile = null;
1692             int per_node_counter = 0;
1693             out.write( "<html>" );
1694             out.write( SurfacingConstants.NL );
1695             addHtmlHead( out, title_for_html );
1696             out.write( SurfacingConstants.NL );
1697             out.write( "<body>" );
1698             out.write( SurfacingConstants.NL );
1699             out.write( "<h1>" );
1700             out.write( SurfacingConstants.NL );
1701             out.write( title_for_html );
1702             out.write( SurfacingConstants.NL );
1703             out.write( "</h1>" );
1704             out.write( SurfacingConstants.NL );
1705             out.write( "<table>" );
1706             out.write( SurfacingConstants.NL );
1707             for( final String id : sorted_ids ) {
1708                 out.write( "<tr>" );
1709                 out.write( "<td>" );
1710                 out.write( "<a href=\"#" + id + "\">" + id + "</a>" );
1711                 writeTaxonomyLinks( out, id );
1712                 out.write( "</td>" );
1713                 out.write( "</tr>" );
1714                 out.write( SurfacingConstants.NL );
1715             }
1716             out.write( "</table>" );
1717             out.write( SurfacingConstants.NL );
1718             for( final String id : sorted_ids ) {
1719                 out.write( SurfacingConstants.NL );
1720                 out.write( "<h2>" );
1721                 out.write( "<a name=\"" + id + "\">" + id + "</a>" );
1722                 writeTaxonomyLinks( out, id );
1723                 out.write( "</h2>" );
1724                 out.write( SurfacingConstants.NL );
1725                 out.write( "<table>" );
1726                 out.write( SurfacingConstants.NL );
1727                 out.write( "<tr>" );
1728                 out.write( "<td><b>" );
1729                 out.write( "Pfam domain(s)" );
1730                 out.write( "</b></td><td><b>" );
1731                 out.write( "GO term acc" );
1732                 out.write( "</b></td><td><b>" );
1733                 out.write( "GO term" );
1734                 out.write( "</b></td><td><b>" );
1735                 out.write( "Penultimate GO term" );
1736                 out.write( "</b></td><td><b>" );
1737                 out.write( "GO namespace" );
1738                 out.write( "</b></td>" );
1739                 out.write( "</tr>" );
1740                 out.write( SurfacingConstants.NL );
1741                 out.write( "</tr>" );
1742                 out.write( SurfacingConstants.NL );
1743                 per_node_counter = 0;
1744                 if ( matrix.getNumberOfCharacters() > 0 ) {
1745                     per_node_go_mapped_domain_gain_loss_outfile = new File( per_node_go_mapped_domain_gain_loss_files_base_dir
1746                             + ForesterUtil.FILE_SEPARATOR + id + suffix_for_per_node_events_file );
1747                     SurfacingUtil.checkForOutputFileWriteability( per_node_go_mapped_domain_gain_loss_outfile );
1748                     per_node_go_mapped_domain_gain_loss_outfile_writer = ForesterUtil
1749                             .createBufferedWriter( per_node_go_mapped_domain_gain_loss_outfile );
1750                 }
1751                 else {
1752                     per_node_go_mapped_domain_gain_loss_outfile = null;
1753                     per_node_go_mapped_domain_gain_loss_outfile_writer = null;
1754                 }
1755                 for( int c = 0; c < matrix.getNumberOfCharacters(); ++c ) {
1756                     // Not nice:
1757                     // using null to indicate either UNCHANGED_PRESENT or GAIN.
1758                     if ( ( matrix.getState( id, c ) == state )
1759                             || ( ( state == null ) && ( ( matrix.getState( id, c ) == CharacterStateMatrix.GainLossStates.UNCHANGED_PRESENT ) || ( matrix
1760                                     .getState( id, c ) == CharacterStateMatrix.GainLossStates.GAIN ) ) ) ) {
1761                         final String character = matrix.getCharacter( c );
1762                         String domain_0 = "";
1763                         String domain_1 = "";
1764                         if ( character.indexOf( BinaryDomainCombination.SEPARATOR ) > 0 ) {
1765                             final String[] s = character.split( BinaryDomainCombination.SEPARATOR );
1766                             if ( s.length != 2 ) {
1767                                 throw new AssertionError( "this should not have happened: unexpected format for domain combination: ["
1768                                         + character + "]" );
1769                             }
1770                             domain_0 = s[ 0 ];
1771                             domain_1 = s[ 1 ];
1772                         }
1773                         else {
1774                             domain_0 = character;
1775                         }
1776                         writeDomainData( domain_id_to_go_ids_map,
1777                                          go_id_to_term_map,
1778                                          go_namespace_limit,
1779                                          out,
1780                                          domain_0,
1781                                          domain_1,
1782                                          prefix_for_html,
1783                                          character_separator,
1784                                          domain_id_to_secondary_features_maps,
1785                                          null );
1786                         all_pfams_encountered.add( domain_0 );
1787                         if ( pfams_gained_or_lost != null ) {
1788                             pfams_gained_or_lost.add( domain_0 );
1789                         }
1790                         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
1791                             all_pfams_encountered.add( domain_1 );
1792                             if ( pfams_gained_or_lost != null ) {
1793                                 pfams_gained_or_lost.add( domain_1 );
1794                             }
1795                         }
1796                         if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1797                             writeDomainsToIndividualFilePerTreeNode( per_node_go_mapped_domain_gain_loss_outfile_writer,
1798                                                                      domain_0,
1799                                                                      domain_1 );
1800                             per_node_counter++;
1801                         }
1802                     }
1803                 }
1804                 if ( per_node_go_mapped_domain_gain_loss_outfile_writer != null ) {
1805                     per_node_go_mapped_domain_gain_loss_outfile_writer.close();
1806                     if ( per_node_counter < 1 ) {
1807                         per_node_go_mapped_domain_gain_loss_outfile.delete();
1808                     }
1809                     per_node_counter = 0;
1810                 }
1811                 out.write( "</table>" );
1812                 out.write( SurfacingConstants.NL );
1813                 out.write( "<hr>" );
1814                 out.write( SurfacingConstants.NL );
1815             } // for( final String id : sorted_ids ) {
1816             out.write( "</body>" );
1817             out.write( SurfacingConstants.NL );
1818             out.write( "</html>" );
1819             out.write( SurfacingConstants.NL );
1820             out.flush();
1821             out.close();
1822         }
1823         catch ( final IOException e ) {
1824             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1825         }
1826         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote characters detailed HTML list: \"" + filename + "\"" );
1827     }
1828
1829     public static void writeDomainCombinationsCountsFile( final String[][] input_file_properties,
1830                                                           final File output_dir,
1831                                                           final Writer per_genome_domain_promiscuity_statistics_writer,
1832                                                           final GenomeWideCombinableDomains gwcd,
1833                                                           final int i,
1834                                                           final GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder dc_sort_order ) {
1835         File dc_outfile = new File( input_file_properties[ i ][ 0 ]
1836                 + surfacing.DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX );
1837         if ( output_dir != null ) {
1838             dc_outfile = new File( output_dir + ForesterUtil.FILE_SEPARATOR + dc_outfile );
1839         }
1840         checkForOutputFileWriteability( dc_outfile );
1841         try {
1842             final BufferedWriter out = new BufferedWriter( new FileWriter( dc_outfile ) );
1843             out.write( gwcd.toStringBuilder( dc_sort_order ).toString() );
1844             out.close();
1845         }
1846         catch ( final IOException e ) {
1847             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1848         }
1849         final DescriptiveStatistics stats = gwcd.getPerGenomeDomainPromiscuityStatistics();
1850         try {
1851             per_genome_domain_promiscuity_statistics_writer.write( input_file_properties[ i ][ 0 ] + "\t" );
1852             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.arithmeticMean() ) + "\t" );
1853             if ( stats.getN() < 2 ) {
1854                 per_genome_domain_promiscuity_statistics_writer.write( "n/a" + "\t" );
1855             }
1856             else {
1857                 per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats
1858                         .sampleStandardDeviation() ) + "\t" );
1859             }
1860             per_genome_domain_promiscuity_statistics_writer.write( FORMATTER_3.format( stats.median() ) + "\t" );
1861             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMin() + "\t" );
1862             per_genome_domain_promiscuity_statistics_writer.write( ( int ) stats.getMax() + "\t" );
1863             per_genome_domain_promiscuity_statistics_writer.write( stats.getN() + "\t" );
1864             final SortedSet<DomainId> mpds = gwcd.getMostPromiscuosDomain();
1865             for( final DomainId mpd : mpds ) {
1866                 per_genome_domain_promiscuity_statistics_writer.write( mpd.getId() + " " );
1867             }
1868             per_genome_domain_promiscuity_statistics_writer.write( ForesterUtil.LINE_SEPARATOR );
1869         }
1870         catch ( final IOException e ) {
1871             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
1872         }
1873         if ( input_file_properties[ i ].length == 3 ) {
1874             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1875                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ", "
1876                     + input_file_properties[ i ][ 2 ] + ") to: \"" + dc_outfile + "\"" );
1877         }
1878         else {
1879             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote domain combination counts for \""
1880                     + input_file_properties[ i ][ 0 ] + "\" (" + input_file_properties[ i ][ 1 ] + ") to: \""
1881                     + dc_outfile + "\"" );
1882         }
1883     }
1884
1885     private static void writeDomainData( final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
1886                                          final Map<GoId, GoTerm> go_id_to_term_map,
1887                                          final GoNameSpace go_namespace_limit,
1888                                          final Writer out,
1889                                          final String domain_0,
1890                                          final String domain_1,
1891                                          final String prefix_for_html,
1892                                          final String character_separator_for_non_html_output,
1893                                          final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps,
1894                                          final Set<GoId> all_go_ids ) throws IOException {
1895         boolean any_go_annotation_present = false;
1896         boolean first_has_no_go = false;
1897         int domain_count = 2; // To distinguish between domains and binary domain combinations.
1898         if ( ForesterUtil.isEmpty( domain_1 ) ) {
1899             domain_count = 1;
1900         }
1901         // The following has a difficult to understand logic.
1902         for( int d = 0; d < domain_count; ++d ) {
1903             List<GoId> go_ids = null;
1904             boolean go_annotation_present = false;
1905             if ( d == 0 ) {
1906                 final DomainId domain_id = new DomainId( domain_0 );
1907                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1908                     go_annotation_present = true;
1909                     any_go_annotation_present = true;
1910                     go_ids = domain_id_to_go_ids_map.get( domain_id );
1911                 }
1912                 else {
1913                     first_has_no_go = true;
1914                 }
1915             }
1916             else {
1917                 final DomainId domain_id = new DomainId( domain_1 );
1918                 if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
1919                     go_annotation_present = true;
1920                     any_go_annotation_present = true;
1921                     go_ids = domain_id_to_go_ids_map.get( domain_id );
1922                 }
1923             }
1924             if ( go_annotation_present ) {
1925                 boolean first = ( ( d == 0 ) || ( ( d == 1 ) && first_has_no_go ) );
1926                 for( final GoId go_id : go_ids ) {
1927                     out.write( "<tr>" );
1928                     if ( first ) {
1929                         first = false;
1930                         writeDomainIdsToHtml( out,
1931                                               domain_0,
1932                                               domain_1,
1933                                               prefix_for_html,
1934                                               domain_id_to_secondary_features_maps );
1935                     }
1936                     else {
1937                         out.write( "<td></td>" );
1938                     }
1939                     if ( !go_id_to_term_map.containsKey( go_id ) ) {
1940                         throw new IllegalArgumentException( "GO-id [" + go_id + "] not found in GO-id to GO-term map" );
1941                     }
1942                     final GoTerm go_term = go_id_to_term_map.get( go_id );
1943                     if ( ( go_namespace_limit == null ) || go_namespace_limit.equals( go_term.getGoNameSpace() ) ) {
1944                         // final String top = GoUtils.getPenultimateGoTerm( go_term, go_id_to_term_map ).getName();
1945                         final String go_id_str = go_id.getId();
1946                         out.write( "<td>" );
1947                         out.write( "<a href=\"" + SurfacingConstants.AMIGO_LINK + go_id_str
1948                                 + "\" target=\"amigo_window\">" + go_id_str + "</a>" );
1949                         out.write( "</td><td>" );
1950                         out.write( go_term.getName() );
1951                         if ( domain_count == 2 ) {
1952                             out.write( " (" + d + ")" );
1953                         }
1954                         out.write( "</td><td>" );
1955                         // out.write( top );
1956                         // out.write( "</td><td>" );
1957                         out.write( "[" );
1958                         out.write( go_term.getGoNameSpace().toShortString() );
1959                         out.write( "]" );
1960                         out.write( "</td>" );
1961                         if ( all_go_ids != null ) {
1962                             all_go_ids.add( go_id );
1963                         }
1964                     }
1965                     else {
1966                         out.write( "<td>" );
1967                         out.write( "</td><td>" );
1968                         out.write( "</td><td>" );
1969                         out.write( "</td><td>" );
1970                         out.write( "</td>" );
1971                     }
1972                     out.write( "</tr>" );
1973                     out.write( SurfacingConstants.NL );
1974                 }
1975             }
1976         } //  for( int d = 0; d < domain_count; ++d )
1977         if ( !any_go_annotation_present ) {
1978             out.write( "<tr>" );
1979             writeDomainIdsToHtml( out, domain_0, domain_1, prefix_for_html, domain_id_to_secondary_features_maps );
1980             out.write( "<td>" );
1981             out.write( "</td><td>" );
1982             out.write( "</td><td>" );
1983             out.write( "</td><td>" );
1984             out.write( "</td>" );
1985             out.write( "</tr>" );
1986             out.write( SurfacingConstants.NL );
1987         }
1988     }
1989
1990     private static void writeDomainIdsToHtml( final Writer out,
1991                                               final String domain_0,
1992                                               final String domain_1,
1993                                               final String prefix_for_detailed_html,
1994                                               final Map<DomainId, Set<String>>[] domain_id_to_secondary_features_maps )
1995             throws IOException {
1996         out.write( "<td>" );
1997         if ( !ForesterUtil.isEmpty( prefix_for_detailed_html ) ) {
1998             out.write( prefix_for_detailed_html );
1999             out.write( " " );
2000         }
2001         out.write( "<a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK + domain_0 + "\">" + domain_0 + "</a>" );
2002         out.write( "</td>" );
2003     }
2004
2005     public static DescriptiveStatistics writeDomainSimilaritiesToFile( final StringBuilder html_desc,
2006                                                                        final StringBuilder html_title,
2007                                                                        final Writer single_writer,
2008                                                                        Map<Character, Writer> split_writers,
2009                                                                        final SortedSet<DomainSimilarity> similarities,
2010                                                                        final boolean treat_as_binary,
2011                                                                        final List<Species> species_order,
2012                                                                        final PrintableDomainSimilarity.PRINT_OPTION print_option,
2013                                                                        final DomainSimilarity.DomainSimilaritySortField sort_field,
2014                                                                        final DomainSimilarity.DomainSimilarityScoring scoring,
2015                                                                        final boolean verbose ) throws IOException {
2016         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
2017         String histogram_title = null;
2018         switch ( sort_field ) {
2019             case ABS_MAX_COUNTS_DIFFERENCE:
2020                 if ( treat_as_binary ) {
2021                     histogram_title = "absolute counts difference:";
2022                 }
2023                 else {
2024                     histogram_title = "absolute (maximal) counts difference:";
2025                 }
2026                 break;
2027             case MAX_COUNTS_DIFFERENCE:
2028                 if ( treat_as_binary ) {
2029                     histogram_title = "counts difference:";
2030                 }
2031                 else {
2032                     histogram_title = "(maximal) counts difference:";
2033                 }
2034                 break;
2035             case DOMAIN_ID:
2036                 histogram_title = "score mean:";
2037                 break;
2038             case MIN:
2039                 histogram_title = "score minimum:";
2040                 break;
2041             case MAX:
2042                 histogram_title = "score maximum:";
2043                 break;
2044             case MAX_DIFFERENCE:
2045                 if ( treat_as_binary ) {
2046                     histogram_title = "difference:";
2047                 }
2048                 else {
2049                     histogram_title = "(maximal) difference:";
2050                 }
2051                 break;
2052             case MEAN:
2053                 histogram_title = "score mean:";
2054                 break;
2055             case SD:
2056                 histogram_title = "score standard deviation:";
2057                 break;
2058             case SPECIES_COUNT:
2059                 histogram_title = "species number:";
2060                 break;
2061             default:
2062                 throw new AssertionError( "Unknown sort field: " + sort_field );
2063         }
2064         for( final DomainSimilarity similarity : similarities ) {
2065             switch ( sort_field ) {
2066                 case ABS_MAX_COUNTS_DIFFERENCE:
2067                     stats.addValue( Math.abs( similarity.getMaximalDifferenceInCounts() ) );
2068                     break;
2069                 case MAX_COUNTS_DIFFERENCE:
2070                     stats.addValue( similarity.getMaximalDifferenceInCounts() );
2071                     break;
2072                 case DOMAIN_ID:
2073                     stats.addValue( similarity.getMeanSimilarityScore() );
2074                     break;
2075                 case MIN:
2076                     stats.addValue( similarity.getMinimalSimilarityScore() );
2077                     break;
2078                 case MAX:
2079                     stats.addValue( similarity.getMaximalSimilarityScore() );
2080                     break;
2081                 case MAX_DIFFERENCE:
2082                     stats.addValue( similarity.getMaximalDifference() );
2083                     break;
2084                 case MEAN:
2085                     stats.addValue( similarity.getMeanSimilarityScore() );
2086                     break;
2087                 case SD:
2088                     stats.addValue( similarity.getStandardDeviationOfSimilarityScore() );
2089                     break;
2090                 case SPECIES_COUNT:
2091                     stats.addValue( similarity.getSpecies().size() );
2092                     break;
2093                 default:
2094                     throw new AssertionError( "Unknown sort field: " + sort_field );
2095             }
2096         }
2097         //
2098         // final HistogramData[] hists = new HistogramData[ 1 ];
2099         //
2100         //
2101         // List<HistogramDataItem> data_items = new
2102         // ArrayList<HistogramDataItem>();
2103         // double[] values = stats.getDataAsDoubleArray();
2104         // for( int i = 0; i < values.length; i++ ) {
2105         // HistogramDataItem data_item = new BasicHistogramDataItem( "", values[
2106         // i ] );
2107         // data_items.add( data_item );
2108         // }
2109         //
2110         //
2111         // HistogramData hd0 = new HistogramData( "name",
2112         // data_items,
2113         // null, 20,
2114         // 40 );
2115         //
2116         //
2117         //
2118         //
2119         // hists[ 0 ] = hd0;
2120         //
2121         // final HistogramsFrame hf = new HistogramsFrame( hists );
2122         // hf.setVisible( true );
2123         //
2124         AsciiHistogram histo = null;
2125         if ( stats.getMin() < stats.getMin() ) {
2126             histo = new AsciiHistogram( stats, histogram_title );
2127         }
2128         if ( verbose ) {
2129             if ( histo != null ) {
2130                 System.out.println( histo.toStringBuffer( 20, '|', 40, 5 ) );
2131             }
2132             System.out.println();
2133             System.out.println( "N                   : " + stats.getN() );
2134             System.out.println( "Min                 : " + stats.getMin() );
2135             System.out.println( "Max                 : " + stats.getMax() );
2136             System.out.println( "Mean                : " + stats.arithmeticMean() );
2137             if ( stats.getN() > 1 ) {
2138                 System.out.println( "SD                  : " + stats.sampleStandardDeviation() );
2139             }
2140             else {
2141                 System.out.println( "SD                  : n/a" );
2142             }
2143             System.out.println( "Median              : " + stats.median() );
2144             if ( stats.getN() > 1 ) {
2145                 System.out.println( "Pearsonian skewness : " + stats.pearsonianSkewness() );
2146             }
2147             else {
2148                 System.out.println( "Pearsonian skewness : n/a" );
2149             }
2150         }
2151         if ( ( single_writer != null ) && ( ( split_writers == null ) || split_writers.isEmpty() ) ) {
2152             split_writers = new HashMap<Character, Writer>();
2153             split_writers.put( '_', single_writer );
2154         }
2155         switch ( print_option ) {
2156             case SIMPLE_TAB_DELIMITED:
2157                 break;
2158             case HTML:
2159                 for( final Character key : split_writers.keySet() ) {
2160                     final Writer w = split_writers.get( key );
2161                     w.write( "<html>" );
2162                     w.write( SurfacingConstants.NL );
2163                     if ( key != '_' ) {
2164                         addHtmlHead( w, "DCs (" + html_title + ") " + key.toString().toUpperCase() );
2165                     }
2166                     else {
2167                         addHtmlHead( w, "DCs (" + html_title + ")" );
2168                     }
2169                     w.write( SurfacingConstants.NL );
2170                     w.write( "<body>" );
2171                     w.write( SurfacingConstants.NL );
2172                     w.write( html_desc.toString() );
2173                     w.write( SurfacingConstants.NL );
2174                     w.write( "<hr>" );
2175                     w.write( "<br>" );
2176                     w.write( SurfacingConstants.NL );
2177                     w.write( "<tt><pre>" );
2178                     w.write( SurfacingConstants.NL );
2179                     if ( histo != null ) {
2180                         w.write( histo.toStringBuffer( 20, '|', 40, 5 ).toString() );
2181                         w.write( SurfacingConstants.NL );
2182                     }
2183                     w.write( "</pre></tt>" );
2184                     w.write( SurfacingConstants.NL );
2185                     w.write( "<table>" );
2186                     w.write( SurfacingConstants.NL );
2187                     w.write( "<tr><td>N: </td><td>" + stats.getN() + "</td></tr>" );
2188                     w.write( SurfacingConstants.NL );
2189                     w.write( "<tr><td>Min: </td><td>" + stats.getMin() + "</td></tr>" );
2190                     w.write( SurfacingConstants.NL );
2191                     w.write( "<tr><td>Max: </td><td>" + stats.getMax() + "</td></tr>" );
2192                     w.write( SurfacingConstants.NL );
2193                     w.write( "<tr><td>Mean: </td><td>" + stats.arithmeticMean() + "</td></tr>" );
2194                     w.write( SurfacingConstants.NL );
2195                     if ( stats.getN() > 1 ) {
2196                         w.write( "<tr><td>SD: </td><td>" + stats.sampleStandardDeviation() + "</td></tr>" );
2197                     }
2198                     else {
2199                         w.write( "<tr><td>SD: </td><td>n/a</td></tr>" );
2200                     }
2201                     w.write( SurfacingConstants.NL );
2202                     w.write( "<tr><td>Median: </td><td>" + stats.median() + "</td></tr>" );
2203                     w.write( SurfacingConstants.NL );
2204                     if ( stats.getN() > 1 ) {
2205                         w.write( "<tr><td>Pearsonian skewness: </td><td>" + stats.pearsonianSkewness() + "</td></tr>" );
2206                     }
2207                     else {
2208                         w.write( "<tr><td>Pearsonian skewness: </td><td>n/a</td></tr>" );
2209                     }
2210                     w.write( SurfacingConstants.NL );
2211                     w.write( "</table>" );
2212                     w.write( SurfacingConstants.NL );
2213                     w.write( "<br>" );
2214                     w.write( SurfacingConstants.NL );
2215                     w.write( "<hr>" );
2216                     w.write( SurfacingConstants.NL );
2217                     w.write( "<br>" );
2218                     w.write( SurfacingConstants.NL );
2219                     w.write( "<table>" );
2220                     w.write( SurfacingConstants.NL );
2221                 }
2222                 break;
2223         }
2224         for( final Writer w : split_writers.values() ) {
2225             w.write( SurfacingConstants.NL );
2226         }
2227         for( final DomainSimilarity similarity : similarities ) {
2228             if ( ( species_order != null ) && !species_order.isEmpty() ) {
2229                 ( ( PrintableDomainSimilarity ) similarity ).setSpeciesOrder( species_order );
2230             }
2231             if ( single_writer != null ) {
2232                 single_writer.write( similarity.toStringBuffer( print_option ).toString() );
2233             }
2234             else {
2235                 Writer local_writer = split_writers.get( ( similarity.getDomainId().getId().charAt( 0 ) + "" )
2236                         .toLowerCase().charAt( 0 ) );
2237                 if ( local_writer == null ) {
2238                     local_writer = split_writers.get( '0' );
2239                 }
2240                 local_writer.write( similarity.toStringBuffer( print_option ).toString() );
2241             }
2242             for( final Writer w : split_writers.values() ) {
2243                 w.write( SurfacingConstants.NL );
2244             }
2245         }
2246         switch ( print_option ) {
2247             case HTML:
2248                 for( final Writer w : split_writers.values() ) {
2249                     w.write( SurfacingConstants.NL );
2250                     w.write( "</table>" );
2251                     w.write( SurfacingConstants.NL );
2252                     w.write( "</font>" );
2253                     w.write( SurfacingConstants.NL );
2254                     w.write( "</body>" );
2255                     w.write( SurfacingConstants.NL );
2256                     w.write( "</html>" );
2257                     w.write( SurfacingConstants.NL );
2258                 }
2259                 break;
2260         }
2261         for( final Writer w : split_writers.values() ) {
2262             w.close();
2263         }
2264         return stats;
2265     }
2266
2267     private static void writeDomainsToIndividualFilePerTreeNode( final Writer individual_files_writer,
2268                                                                  final String domain_0,
2269                                                                  final String domain_1 ) throws IOException {
2270         individual_files_writer.write( domain_0 );
2271         individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2272         if ( !ForesterUtil.isEmpty( domain_1 ) ) {
2273             individual_files_writer.write( domain_1 );
2274             individual_files_writer.write( ForesterUtil.LINE_SEPARATOR );
2275         }
2276     }
2277
2278     public static void writeMatrixToFile( final CharacterStateMatrix<?> matrix,
2279                                           final String filename,
2280                                           final Format format ) {
2281         final File outfile = new File( filename );
2282         checkForOutputFileWriteability( outfile );
2283         try {
2284             final BufferedWriter out = new BufferedWriter( new FileWriter( outfile ) );
2285             matrix.toWriter( out, format );
2286             out.flush();
2287             out.close();
2288         }
2289         catch ( final IOException e ) {
2290             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2291         }
2292         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote matrix: \"" + filename + "\"" );
2293     }
2294
2295     public static void writeMatrixToFile( final File matrix_outfile, final List<DistanceMatrix> matrices ) {
2296         checkForOutputFileWriteability( matrix_outfile );
2297         try {
2298             final BufferedWriter out = new BufferedWriter( new FileWriter( matrix_outfile ) );
2299             for( final DistanceMatrix distance_matrix : matrices ) {
2300                 out.write( distance_matrix.toStringBuffer( DistanceMatrix.Format.PHYLIP ).toString() );
2301                 out.write( ForesterUtil.LINE_SEPARATOR );
2302                 out.flush();
2303             }
2304             out.close();
2305         }
2306         catch ( final IOException e ) {
2307             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2308         }
2309         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote distance matrices to \"" + matrix_outfile + "\"" );
2310     }
2311
2312     private static void writePfamsToFile( final String outfile_name, final SortedSet<String> pfams ) {
2313         try {
2314             final Writer writer = new BufferedWriter( new FileWriter( new File( outfile_name ) ) );
2315             for( final String pfam : pfams ) {
2316                 writer.write( pfam );
2317                 writer.write( ForesterUtil.LINE_SEPARATOR );
2318             }
2319             writer.close();
2320             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote " + pfams.size() + " pfams to [" + outfile_name
2321                     + "]" );
2322         }
2323         catch ( final IOException e ) {
2324             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "Failure to write: " + e );
2325         }
2326     }
2327
2328     public static void writePhylogenyToFile( final Phylogeny phylogeny, final String filename ) {
2329         final PhylogenyWriter writer = new PhylogenyWriter();
2330         try {
2331             writer.toPhyloXML( new File( filename ), phylogeny, 1 );
2332         }
2333         catch ( final IOException e ) {
2334             ForesterUtil.printWarningMessage( surfacing.PRG_NAME, "failed to write phylogeny to \"" + filename + "\": "
2335                     + e );
2336         }
2337         ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote phylogeny to \"" + filename + "\"" );
2338     }
2339
2340     public static void writeTaxonomyLinks( final Writer writer, final String species ) throws IOException {
2341         if ( ( species.length() > 1 ) && ( species.indexOf( '_' ) < 1 ) ) {
2342             final Matcher matcher = PATTERN_SP_STYLE_TAXONOMY.matcher( species );
2343             writer.write( " [" );
2344             if ( matcher.matches() ) {
2345                 writer.write( "<a href=\"" + SurfacingConstants.UNIPROT_LINK + species
2346                         + "\" target=\"taxonomy_window\">uniprot</a>" );
2347             }
2348             else {
2349                 writer.write( "<a href=\"" + SurfacingConstants.EOL_LINK + species
2350                         + "\" target=\"taxonomy_window\">eol</a>" );
2351                 writer.write( "|" );
2352                 writer.write( "<a href=\"" + SurfacingConstants.TOL_LINK + species
2353                         + "\" target=\"taxonomy_window\">tol</a>" );
2354             }
2355             writer.write( "]" );
2356         }
2357     }
2358
2359     private static void writeToNexus( final String outfile_name,
2360                                       final CharacterStateMatrix<BinaryStates> matrix,
2361                                       final Phylogeny phylogeny ) {
2362         if ( !( matrix instanceof BasicCharacterStateMatrix ) ) {
2363             throw new IllegalArgumentException( "can only write matrices of type [" + BasicCharacterStateMatrix.class
2364                     + "] to nexus" );
2365         }
2366         final BasicCharacterStateMatrix<BinaryStates> my_matrix = ( org.forester.evoinference.matrix.character.BasicCharacterStateMatrix<BinaryStates> ) matrix;
2367         final List<Phylogeny> phylogenies = new ArrayList<Phylogeny>( 1 );
2368         phylogenies.add( phylogeny );
2369         try {
2370             final BufferedWriter w = new BufferedWriter( new FileWriter( outfile_name ) );
2371             w.write( NexusConstants.NEXUS );
2372             w.write( ForesterUtil.LINE_SEPARATOR );
2373             my_matrix.writeNexusTaxaBlock( w );
2374             my_matrix.writeNexusBinaryChractersBlock( w );
2375             PhylogenyWriter.writeNexusTreesBlock( w, phylogenies, NH_CONVERSION_SUPPORT_VALUE_STYLE.NONE );
2376             w.flush();
2377             w.close();
2378             ForesterUtil.programMessage( surfacing.PRG_NAME, "Wrote Nexus file: \"" + outfile_name + "\"" );
2379         }
2380         catch ( final IOException e ) {
2381             ForesterUtil.fatalError( surfacing.PRG_NAME, e.getMessage() );
2382         }
2383     }
2384
2385     private static void writeToNexus( final String outfile_name,
2386                                       final DomainParsimonyCalculator domain_parsimony,
2387                                       final Phylogeny phylogeny ) {
2388         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAINS,
2389                       domain_parsimony.createMatrixOfDomainPresenceOrAbsence(),
2390                       phylogeny );
2391         writeToNexus( outfile_name + surfacing.NEXUS_EXTERNAL_DOMAIN_COMBINATIONS,
2392                       domain_parsimony.createMatrixOfBinaryDomainCombinationPresenceOrAbsence(),
2393                       phylogeny );
2394     }
2395
2396     public static void domainsPerProteinsStatistics( final String genome,
2397                                                      final List<Protein> protein_list,
2398                                                      final DescriptiveStatistics all_genomes_domains_per_potein_stats,
2399                                                      final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
2400                                                      final SortedSet<String> domains_which_are_always_single,
2401                                                      final SortedSet<String> domains_which_are_sometimes_single_sometimes_not,
2402                                                      final SortedSet<String> domains_which_never_single,
2403                                                      final Writer writer ) {
2404         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
2405         for( final Protein protein : protein_list ) {
2406             final int domains = protein.getNumberOfProteinDomains();
2407             //System.out.println( domains );
2408             stats.addValue( domains );
2409             all_genomes_domains_per_potein_stats.addValue( domains );
2410             if ( !all_genomes_domains_per_potein_histo.containsKey( domains ) ) {
2411                 all_genomes_domains_per_potein_histo.put( domains, 1 );
2412             }
2413             else {
2414                 all_genomes_domains_per_potein_histo.put( domains,
2415                                                           1 + all_genomes_domains_per_potein_histo.get( domains ) );
2416             }
2417             if ( domains == 1 ) {
2418                 final String domain = protein.getProteinDomain( 0 ).getDomainId().getId();
2419                 if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
2420                     if ( domains_which_never_single.contains( domain ) ) {
2421                         domains_which_never_single.remove( domain );
2422                         domains_which_are_sometimes_single_sometimes_not.add( domain );
2423                     }
2424                     else {
2425                         domains_which_are_always_single.add( domain );
2426                     }
2427                 }
2428             }
2429             else if ( domains > 1 ) {
2430                 for( final Domain d : protein.getProteinDomains() ) {
2431                     final String domain = d.getDomainId().getId();
2432                     // System.out.println( domain );
2433                     if ( !domains_which_are_sometimes_single_sometimes_not.contains( domain ) ) {
2434                         if ( domains_which_are_always_single.contains( domain ) ) {
2435                             domains_which_are_always_single.remove( domain );
2436                             domains_which_are_sometimes_single_sometimes_not.add( domain );
2437                         }
2438                         else {
2439                             domains_which_never_single.add( domain );
2440                         }
2441                     }
2442                 }
2443             }
2444         }
2445         try {
2446             writer.write( genome );
2447             writer.write( "\t" );
2448             if ( stats.getN() >= 1 ) {
2449                 writer.write( stats.arithmeticMean() + "" );
2450                 writer.write( "\t" );
2451                 if ( stats.getN() >= 2 ) {
2452                     writer.write( stats.sampleStandardDeviation() + "" );
2453                 }
2454                 else {
2455                     writer.write( "" );
2456                 }
2457                 writer.write( "\t" );
2458                 writer.write( stats.median() + "" );
2459                 writer.write( "\t" );
2460                 writer.write( stats.getN() + "" );
2461                 writer.write( "\t" );
2462                 writer.write( stats.getMin() + "" );
2463                 writer.write( "\t" );
2464                 writer.write( stats.getMax() + "" );
2465             }
2466             else {
2467                 writer.write( "\t" );
2468                 writer.write( "\t" );
2469                 writer.write( "\t" );
2470                 writer.write( "0" );
2471                 writer.write( "\t" );
2472                 writer.write( "\t" );
2473             }
2474             writer.write( "\n" );
2475         }
2476         catch ( final IOException e ) {
2477             e.printStackTrace();
2478         }
2479     }
2480 }