forester/java/src/org/forester/surfacing/DomainCountsDifferenceUtil.java

   1 // $Id:
   2 // $
   3 //
   4 // FORESTER -- software libraries and applications
   5 // for evolutionary biology research and applications.
   6 //
   7 // Copyright (C) 2008-2009 Christian M. Zmasek
   8 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   9 // All rights reserved
  10 //
  11 // This library is free software; you can redistribute it and/or
  12 // modify it under the terms of the GNU Lesser General Public
  13 // License as published by the Free Software Foundation; either
  14 // version 2.1 of the License, or (at your option) any later version.
  15 //
  16 // This library is distributed in the hope that it will be useful,
  17 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  19 // Lesser General Public License for more details.
  20 //
  21 // You should have received a copy of the GNU Lesser General Public
  22 // License along with this library; if not, write to the Free Software
  23 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  24 //
  25 // Contact: phylosoft @ gmail . com
  26 // WWW: www.phylosoft.org/forester
  27
  28 package org.forester.surfacing;
  29
  30 import java.io.BufferedWriter;
  31 import java.io.File;
  32 import java.io.FileWriter;
  33 import java.io.IOException;
  34 import java.io.Writer;
  35 import java.text.DecimalFormat;
  36 import java.text.NumberFormat;
  37 import java.util.ArrayList;
  38 import java.util.HashMap;
  39 import java.util.List;
  40 import java.util.Map;
  41 import java.util.Set;
  42 import java.util.SortedMap;
  43 import java.util.SortedSet;
  44 import java.util.TreeMap;
  45 import java.util.TreeSet;
  46
  47 import org.forester.application.surfacing;
  48 import org.forester.go.GoId;
  49 import org.forester.go.GoTerm;
  50 import org.forester.util.BasicDescriptiveStatistics;
  51 import org.forester.util.DescriptiveStatistics;
  52 import org.forester.util.ForesterUtil;
  53
  54 /*
  55  * Poorly designed static class which essential has one method:
  56  * calculateCopyNumberDifferences.
  57  */
  58 public final class DomainCountsDifferenceUtil {
  59
  60     private final static NumberFormat          FORMATTER                                   = new DecimalFormat( "0.0E0" );
  61     private static final COPY_CALCULATION_MODE COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES = COPY_CALCULATION_MODE.MIN;
  62     private static final COPY_CALCULATION_MODE COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES   = COPY_CALCULATION_MODE.MIN;
  63     private static final COPY_CALCULATION_MODE COPY_CALC_MODE_FOR_LOW_COPY_SPECIES         = COPY_CALCULATION_MODE.MAX;
  64     private static final String                PLUS_MINUS_PROTEINS_FILE_DOM_SUFFIX         = ".prot";
  65
  66     //FIXME really needs to be tested!
  67     private static void addCounts( final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
  68                                    final BinaryDomainCombination dc,
  69                                    final GenomeWideCombinableDomains genome,
  70                                    final Set<BinaryDomainCombination> bdc ) {
  71         if ( !copy_counts.containsKey( dc ) ) {
  72             copy_counts.put( dc, new ArrayList<Integer>() );
  73         }
  74         if ( bdc.contains( dc )
  75                 && ( ( ( BasicCombinableDomains ) genome.get( dc.getId0() ) ).getCombiningDomains().get( dc.getId1() ) != null ) ) {
  76             final int count = ( ( BasicCombinableDomains ) genome.get( dc.getId0() ) ).getCombiningDomains()
  77                     .get( dc.getId1() );
  78             copy_counts.get( dc ).add( count );
  79         }
  80         else {
  81             copy_counts.get( dc ).add( 0 );
  82         }
  83     }
  84
  85     private static void addCounts( final SortedMap<DomainId, List<Integer>> copy_counts,
  86                                    final DomainId domain,
  87                                    final GenomeWideCombinableDomains genome ) {
  88         if ( !copy_counts.containsKey( domain ) ) {
  89             copy_counts.put( domain, new ArrayList<Integer>() );
  90         }
  91         if ( genome.contains( domain ) ) {
  92             copy_counts.get( domain ).add( genome.get( domain ).getKeyDomainProteinsCount() );
  93         }
  94         else {
  95             copy_counts.get( domain ).add( 0 );
  96         }
  97     }
  98
  99     private static StringBuilder addGoInformation( final DomainId d,
 100                                                    final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 101                                                    final Map<GoId, GoTerm> go_id_to_term_map ) {
 102         final StringBuilder sb = new StringBuilder();
 103         if ( ( domain_id_to_go_ids_map == null ) || domain_id_to_go_ids_map.isEmpty()
 104                 || !domain_id_to_go_ids_map.containsKey( d ) ) {
 105             return sb;
 106         }
 107         final List<GoId> go_ids = domain_id_to_go_ids_map.get( d );
 108         for( int i = 0; i < go_ids.size(); ++i ) {
 109             final GoId go_id = go_ids.get( i );
 110             if ( go_id_to_term_map.containsKey( go_id ) ) {
 111                 appendGoTerm( sb, go_id_to_term_map.get( go_id ) );
 112                 sb.append( "<br>" );
 113             }
 114             else {
 115                 sb.append( "go id \"" + go_id + "\" not found [" + d.getId() + "]" );
 116             }
 117         }
 118         return sb;
 119     }
 120
 121     private static void appendGoTerm( final StringBuilder sb, final GoTerm go_term ) {
 122         final GoId go_id = go_term.getGoId();
 123         sb.append( "<a href=\"" + SurfacingConstants.AMIGO_LINK + go_id + "\" target=\"amigo_window\">" + go_id
 124                 + "</a>" );
 125         sb.append( ":" );
 126         sb.append( go_term.getName() );
 127         sb.append( " [" );
 128         sb.append( go_term.getGoNameSpace().toShortString() );
 129         sb.append( "]" );
 130     }
 131
 132     public static void calculateCopyNumberDifferences( final List<GenomeWideCombinableDomains> genomes,
 133                                                        final SortedMap<Species, List<Protein>> protein_lists_per_species,
 134                                                        final List<String> high_copy_base_species,
 135                                                        final List<String> high_copy_target_species,
 136                                                        final List<String> low_copy_species,
 137                                                        final int min_diff,
 138                                                        final Double factor,
 139                                                        final File plain_output_dom,
 140                                                        final File html_output_dom,
 141                                                        final File html_output_dc,
 142                                                        final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 143                                                        final Map<GoId, GoTerm> go_id_to_term_map,
 144                                                        final File all_domains_go_ids_out_dom,
 145                                                        final File passing_domains_go_ids_out_dom,
 146                                                        final File proteins_file_base ) throws IOException {
 147         if ( genomes.size() < 1 ) {
 148             throw new IllegalArgumentException( "attempt to use empty list of genomes for domain difference calculation" );
 149         }
 150         if ( ( high_copy_base_species.size() < 1 ) || ( low_copy_species.size() < 1 ) ) {
 151             throw new IllegalArgumentException( "attempt to use empty list of species for domain difference calculation" );
 152         }
 153         if ( high_copy_base_species.contains( high_copy_target_species )
 154                 || low_copy_species.contains( high_copy_target_species ) ) {
 155             throw new IllegalArgumentException( "species [" + high_copy_target_species
 156                     + "] appears in other list as well" );
 157         }
 158         if ( min_diff < 0 ) {
 159             throw new IllegalArgumentException( "attempt to use negative addition [" + min_diff + "]" );
 160         }
 161         if ( factor <= 0.0 ) {
 162             throw new IllegalArgumentException( "attempt to use factor equal or smaller than 0.0 [" + factor + "]" );
 163         }
 164         SurfacingUtil.checkForOutputFileWriteability( plain_output_dom );
 165         SurfacingUtil.checkForOutputFileWriteability( html_output_dom );
 166         SurfacingUtil.checkForOutputFileWriteability( html_output_dc );
 167         SurfacingUtil.checkForOutputFileWriteability( all_domains_go_ids_out_dom );
 168         SurfacingUtil.checkForOutputFileWriteability( passing_domains_go_ids_out_dom );
 169         final Writer plain_writer = new BufferedWriter( new FileWriter( plain_output_dom ) );
 170         final Writer html_writer = new BufferedWriter( new FileWriter( html_output_dom ) );
 171         final Writer html_writer_dc = new BufferedWriter( new FileWriter( html_output_dc ) );
 172         final Writer all_gos_writer = new BufferedWriter( new FileWriter( all_domains_go_ids_out_dom ) );
 173         final Writer passing_gos_writer = new BufferedWriter( new FileWriter( passing_domains_go_ids_out_dom ) );
 174         final SortedMap<DomainId, Double> high_copy_base_values = new TreeMap<DomainId, Double>();
 175         final SortedMap<DomainId, Double> high_copy_target_values = new TreeMap<DomainId, Double>();
 176         final SortedMap<DomainId, Double> low_copy_values = new TreeMap<DomainId, Double>();
 177         final SortedMap<DomainId, List<Integer>> high_copy_base_copy_counts = new TreeMap<DomainId, List<Integer>>();
 178         final SortedMap<DomainId, List<Integer>> high_copy_target_copy_counts = new TreeMap<DomainId, List<Integer>>();
 179         final SortedMap<DomainId, List<Integer>> low_copy_copy_counts = new TreeMap<DomainId, List<Integer>>();
 180         final SortedSet<DomainId> all_domains = new TreeSet<DomainId>();
 181         final SortedMap<BinaryDomainCombination, Double> high_copy_base_values_dc = new TreeMap<BinaryDomainCombination, Double>();
 182         final SortedMap<BinaryDomainCombination, Double> high_copy_target_values_dc = new TreeMap<BinaryDomainCombination, Double>();
 183         final SortedMap<BinaryDomainCombination, Double> low_copy_values_dc = new TreeMap<BinaryDomainCombination, Double>();
 184         final SortedMap<BinaryDomainCombination, List<Integer>> high_copy_base_copy_counts_dc = new TreeMap<BinaryDomainCombination, List<Integer>>();
 185         final SortedMap<BinaryDomainCombination, List<Integer>> high_copy_target_copy_counts_dc = new TreeMap<BinaryDomainCombination, List<Integer>>();
 186         final SortedMap<BinaryDomainCombination, List<Integer>> low_copy_copy_counts_dc = new TreeMap<BinaryDomainCombination, List<Integer>>();
 187         final SortedSet<BinaryDomainCombination> all_dcs = new TreeSet<BinaryDomainCombination>();
 188         final Map<String, Set<BinaryDomainCombination>> bdcs_per_genome = new HashMap<String, Set<BinaryDomainCombination>>();
 189         final SortedSet<GoId> go_ids_of_passing_domains = new TreeSet<GoId>();
 190         final SortedSet<GoId> go_ids_all = new TreeSet<GoId>();
 191         for( final GenomeWideCombinableDomains genome : genomes ) {
 192             final SortedSet<DomainId> domains = genome.getAllDomainIds();
 193             final SortedSet<BinaryDomainCombination> dcs = genome.toBinaryDomainCombinations();
 194             final String species = genome.getSpecies().getSpeciesId();
 195             bdcs_per_genome.put( species, genome.toBinaryDomainCombinations() );
 196             for( final DomainId d : domains ) {
 197                 all_domains.add( d );
 198                 if ( domain_id_to_go_ids_map.containsKey( d ) ) {
 199                     go_ids_all.addAll( domain_id_to_go_ids_map.get( d ) );
 200                 }
 201             }
 202             for( final BinaryDomainCombination dc : dcs ) {
 203                 all_dcs.add( dc );
 204             }
 205         }
 206         for( final DomainId domain : all_domains ) {
 207             for( final GenomeWideCombinableDomains genome : genomes ) {
 208                 final String species = genome.getSpecies().getSpeciesId();
 209                 if ( high_copy_base_species.contains( species ) ) {
 210                     DomainCountsDifferenceUtil.addCounts( high_copy_base_copy_counts, domain, genome );
 211                 }
 212                 if ( high_copy_target_species.contains( species ) ) {
 213                     DomainCountsDifferenceUtil.addCounts( high_copy_target_copy_counts, domain, genome );
 214                 }
 215                 if ( low_copy_species.contains( species ) ) {
 216                     DomainCountsDifferenceUtil.addCounts( low_copy_copy_counts, domain, genome );
 217                 }
 218             }
 219         }
 220         for( final BinaryDomainCombination dc : all_dcs ) {
 221             for( final GenomeWideCombinableDomains genome : genomes ) {
 222                 final String species = genome.getSpecies().getSpeciesId();
 223                 if ( high_copy_base_species.contains( species ) ) {
 224                     DomainCountsDifferenceUtil.addCounts( high_copy_base_copy_counts_dc,
 225                                                           dc,
 226                                                           genome,
 227                                                           bdcs_per_genome.get( species ) );
 228                 }
 229                 if ( high_copy_target_species.contains( species ) ) {
 230                     DomainCountsDifferenceUtil.addCounts( high_copy_target_copy_counts_dc,
 231                                                           dc,
 232                                                           genome,
 233                                                           bdcs_per_genome.get( species ) );
 234                 }
 235                 if ( low_copy_species.contains( species ) ) {
 236                     DomainCountsDifferenceUtil.addCounts( low_copy_copy_counts_dc,
 237                                                           dc,
 238                                                           genome,
 239                                                           bdcs_per_genome.get( species ) );
 240                 }
 241             }
 242         }
 243         for( final DomainId domain : all_domains ) {
 244             calculateDomainCountsBasedValue( high_copy_target_values,
 245                                              high_copy_target_copy_counts,
 246                                              domain,
 247                                              COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES );
 248             calculateDomainCountsBasedValue( high_copy_base_values,
 249                                              high_copy_base_copy_counts,
 250                                              domain,
 251                                              COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES );
 252             calculateDomainCountsBasedValue( low_copy_values,
 253                                              low_copy_copy_counts,
 254                                              domain,
 255                                              COPY_CALC_MODE_FOR_LOW_COPY_SPECIES );
 256         }
 257         for( final BinaryDomainCombination dc : all_dcs ) {
 258             calculateDomainCountsBasedValue( high_copy_target_values_dc,
 259                                              high_copy_target_copy_counts_dc,
 260                                              dc,
 261                                              COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES );
 262             calculateDomainCountsBasedValue( high_copy_base_values_dc,
 263                                              high_copy_base_copy_counts_dc,
 264                                              dc,
 265                                              COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES );
 266             calculateDomainCountsBasedValue( low_copy_values_dc,
 267                                              low_copy_copy_counts_dc,
 268                                              dc,
 269                                              COPY_CALC_MODE_FOR_LOW_COPY_SPECIES );
 270         }
 271         writeDomainValuesToFiles( genomes,
 272                                   high_copy_base_species,
 273                                   high_copy_target_species,
 274                                   low_copy_species,
 275                                   min_diff,
 276                                   factor,
 277                                   domain_id_to_go_ids_map,
 278                                   go_id_to_term_map,
 279                                   plain_writer,
 280                                   html_writer,
 281                                   proteins_file_base,
 282                                   high_copy_base_values,
 283                                   high_copy_target_values,
 284                                   low_copy_values,
 285                                   all_domains,
 286                                   go_ids_of_passing_domains,
 287                                   protein_lists_per_species );
 288         writeDomainCombinationValuesToFiles( genomes,
 289                                              high_copy_base_species,
 290                                              high_copy_target_species,
 291                                              low_copy_species,
 292                                              min_diff,
 293                                              factor,
 294                                              html_writer_dc,
 295                                              high_copy_base_values_dc,
 296                                              high_copy_target_values_dc,
 297                                              low_copy_values_dc,
 298                                              all_dcs,
 299                                              bdcs_per_genome );
 300         writeGoIdsToFile( all_gos_writer, go_ids_all );
 301         writeGoIdsToFile( passing_gos_writer, go_ids_of_passing_domains );
 302     }
 303
 304     private static void calculateDomainCountsBasedValue( final SortedMap<BinaryDomainCombination, Double> copy_values,
 305                                                          final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
 306                                                          final BinaryDomainCombination bdc,
 307                                                          final COPY_CALCULATION_MODE copy_calc_mode ) {
 308         if ( copy_counts.containsKey( bdc ) ) {
 309             switch ( copy_calc_mode ) {
 310                 case MAX:
 311                     DomainCountsDifferenceUtil.calculateMaxCount( copy_values, copy_counts, bdc );
 312                     break;
 313                 case MIN:
 314                     DomainCountsDifferenceUtil.calculateMinCount( copy_values, copy_counts, bdc );
 315                     break;
 316                 case MEAN:
 317                     DomainCountsDifferenceUtil.calculateMeanCount( copy_values, copy_counts, bdc );
 318                     break;
 319                 case MEDIAN:
 320                     DomainCountsDifferenceUtil.calculateMedianCount( copy_values, copy_counts, bdc );
 321                     break;
 322                 default:
 323                     throw new IllegalArgumentException();
 324             }
 325         }
 326         else {
 327             copy_values.put( bdc, Double.valueOf( 0.0 ) );
 328         }
 329     }
 330
 331     private static void calculateDomainCountsBasedValue( final SortedMap<DomainId, Double> copy_values,
 332                                                          final SortedMap<DomainId, List<Integer>> copy_counts,
 333                                                          final DomainId domain,
 334                                                          final COPY_CALCULATION_MODE copy_calc_mode ) {
 335         if ( copy_counts.containsKey( domain ) ) {
 336             switch ( copy_calc_mode ) {
 337                 case MAX:
 338                     DomainCountsDifferenceUtil.calculateMaxCount( copy_values, copy_counts, domain );
 339                     break;
 340                 case MIN:
 341                     DomainCountsDifferenceUtil.calculateMinCount( copy_values, copy_counts, domain );
 342                     break;
 343                 case MEAN:
 344                     DomainCountsDifferenceUtil.calculateMeanCount( copy_values, copy_counts, domain );
 345                     break;
 346                 case MEDIAN:
 347                     DomainCountsDifferenceUtil.calculateMedianCount( copy_values, copy_counts, domain );
 348                     break;
 349                 default:
 350                     throw new IllegalArgumentException();
 351             }
 352         }
 353         else {
 354             copy_values.put( domain, Double.valueOf( 0.0 ) );
 355         }
 356     }
 357
 358     private static void calculateMaxCount( final SortedMap<BinaryDomainCombination, Double> results,
 359                                            final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
 360                                            final BinaryDomainCombination bdc ) {
 361         final List<Integer> counts = copy_counts.get( bdc );
 362         int max = 0;
 363         for( final Integer count : counts ) {
 364             if ( count > max ) {
 365                 max = count;
 366             }
 367         }
 368         results.put( bdc, ( double ) max );
 369     }
 370
 371     private static void calculateMaxCount( final SortedMap<DomainId, Double> results,
 372                                            final SortedMap<DomainId, List<Integer>> copy_counts,
 373                                            final DomainId domain ) {
 374         final List<Integer> counts = copy_counts.get( domain );
 375         int max = 0;
 376         for( final Integer count : counts ) {
 377             if ( count > max ) {
 378                 max = count;
 379             }
 380         }
 381         results.put( domain, ( double ) max );
 382     }
 383
 384     private static void calculateMeanCount( final SortedMap<BinaryDomainCombination, Double> results,
 385                                             final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
 386                                             final BinaryDomainCombination bdc ) {
 387         final List<Integer> counts = copy_counts.get( bdc );
 388         int sum = 0;
 389         for( final Integer count : counts ) {
 390             sum += count;
 391         }
 392         results.put( bdc, ( ( double ) sum ) / ( ( double ) counts.size() ) );
 393     }
 394
 395     private static void calculateMeanCount( final SortedMap<DomainId, Double> results,
 396                                             final SortedMap<DomainId, List<Integer>> copy_counts,
 397                                             final DomainId domain ) {
 398         final List<Integer> counts = copy_counts.get( domain );
 399         int sum = 0;
 400         for( final Integer count : counts ) {
 401             sum += count;
 402         }
 403         results.put( domain, ( ( double ) sum ) / ( ( double ) counts.size() ) );
 404     }
 405
 406     private static void calculateMedianCount( final SortedMap<BinaryDomainCombination, Double> results,
 407                                               final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
 408                                               final BinaryDomainCombination bdc ) {
 409         final List<Integer> counts = copy_counts.get( bdc );
 410         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 411         for( final Integer count : counts ) {
 412             stats.addValue( count );
 413         }
 414         results.put( bdc, stats.median() );
 415     }
 416
 417     private static void calculateMedianCount( final SortedMap<DomainId, Double> results,
 418                                               final SortedMap<DomainId, List<Integer>> copy_counts,
 419                                               final DomainId domain ) {
 420         final List<Integer> counts = copy_counts.get( domain );
 421         final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
 422         for( final Integer count : counts ) {
 423             stats.addValue( count );
 424         }
 425         results.put( domain, stats.median() );
 426     }
 427
 428     private static void calculateMinCount( final SortedMap<BinaryDomainCombination, Double> results,
 429                                            final SortedMap<BinaryDomainCombination, List<Integer>> copy_counts,
 430                                            final BinaryDomainCombination bdc ) {
 431         final List<Integer> counts = copy_counts.get( bdc );
 432         int min = Integer.MAX_VALUE;
 433         for( final Integer count : counts ) {
 434             if ( count < min ) {
 435                 min = count;
 436             }
 437         }
 438         results.put( bdc, ( double ) min );
 439     }
 440
 441     private static void calculateMinCount( final SortedMap<DomainId, Double> results,
 442                                            final SortedMap<DomainId, List<Integer>> copy_counts,
 443                                            final DomainId domain ) {
 444         final List<Integer> counts = copy_counts.get( domain );
 445         int min = Integer.MAX_VALUE;
 446         for( final Integer count : counts ) {
 447             if ( count < min ) {
 448                 min = count;
 449             }
 450         }
 451         results.put( domain, ( double ) min );
 452     }
 453
 454     private static String combinableDomaindToString( final CombinableDomains cd ) {
 455         final StringBuilder sb = new StringBuilder();
 456         sb.append( cd.getKeyDomainProteinsCount() );
 457         sb.append( "\t[" );
 458         sb.append( FORMATTER.format( cd.getKeyDomainConfidenceDescriptiveStatistics().median() ) );
 459         sb.append( "]" );
 460         return sb.toString();
 461     }
 462
 463     private static String combinableDomaindToStringHtml( final CombinableDomains cd ) {
 464         final StringBuilder sb = new StringBuilder();
 465         sb.append( "[" );
 466         sb.append( cd.getKeyDomainCount() );
 467         sb.append( ", <b>" );
 468         sb.append( cd.getKeyDomainProteinsCount() );
 469         sb.append( "</b>, " );
 470         sb.append( cd.getNumberOfCombinableDomains() );
 471         sb.append( "]</td><td>[" );
 472         sb.append( FORMATTER.format( cd.getKeyDomainConfidenceDescriptiveStatistics().median() ) );
 473         sb.append( "]</td><td>" );
 474         sb.append( cd.getCombiningDomainIdsAsStringBuilder() );
 475         return sb.toString();
 476     }
 477
 478     private static void writeCopyNumberValues( final SortedMap<BinaryDomainCombination, Double> copy_means,
 479                                                final BinaryDomainCombination bdc,
 480                                                final GenomeWideCombinableDomains genome,
 481                                                final Map<String, Set<BinaryDomainCombination>> bdcs_per_genome,
 482                                                final String species,
 483                                                final Writer html_writer,
 484                                                final String color ) throws IOException {
 485         html_writer.write( "<td> " );
 486         if ( !ForesterUtil.isEmpty( color ) ) {
 487             html_writer.write( "<font color=\"" + color + "\">" );
 488         }
 489         html_writer.write( "<b>" + species + ":</b> " );
 490         if ( !ForesterUtil.isEmpty( color ) ) {
 491             html_writer.write( "</font>" );
 492         }
 493         html_writer.write( "</td><td>" );
 494         if ( bdcs_per_genome.get( species ).contains( bdc ) && ( copy_means.get( bdc ) > 0 ) ) {
 495             final int count = ( ( BasicCombinableDomains ) genome.get( bdc.getId0() ) ).getCombiningDomains()
 496                     .get( bdc.getId1() );
 497             html_writer.write( count + "" );
 498         }
 499         else {
 500             html_writer.write( "0" );
 501         }
 502         html_writer.write( "</td>" );
 503     }
 504
 505     private static void writeCopyNumberValues( final SortedMap<DomainId, Double> copy_means,
 506                                                final DomainId domain,
 507                                                final GenomeWideCombinableDomains genome,
 508                                                final String species,
 509                                                final Writer plain_writer,
 510                                                final Writer html_writer,
 511                                                final String color ) throws IOException {
 512         plain_writer.write( "  " + species + "\t" );
 513         html_writer.write( "<td> " );
 514         if ( !ForesterUtil.isEmpty( color ) ) {
 515             html_writer.write( "<font color=\"" + color + "\">" );
 516         }
 517         html_writer.write( "<b>" + species + ":</b> " );
 518         if ( !ForesterUtil.isEmpty( color ) ) {
 519             html_writer.write( "</font>" );
 520         }
 521         html_writer.write( "</td><td>" );
 522         if ( genome.contains( domain ) && ( copy_means.get( domain ) > 0 ) ) {
 523             plain_writer.write( DomainCountsDifferenceUtil.combinableDomaindToString( genome.get( domain ) ) );
 524             html_writer.write( DomainCountsDifferenceUtil.combinableDomaindToStringHtml( genome.get( domain ) ) );
 525         }
 526         else {
 527             plain_writer.write( "0" );
 528             html_writer.write( "0" );
 529         }
 530         html_writer.write( "</td>" );
 531         plain_writer.write( SurfacingConstants.NL );
 532     }
 533
 534     private static void writeDomainCombinationValuesToFiles( final List<GenomeWideCombinableDomains> genomes,
 535                                                              final List<String> high_copy_base_species,
 536                                                              final List<String> high_copy_target_species,
 537                                                              final List<String> low_copy_species,
 538                                                              final int min_diff,
 539                                                              final Double factor,
 540                                                              final Writer html_writer,
 541                                                              final SortedMap<BinaryDomainCombination, Double> high_copy_base_values,
 542                                                              final SortedMap<BinaryDomainCombination, Double> high_copy_target_values,
 543                                                              final SortedMap<BinaryDomainCombination, Double> low_copy_values,
 544                                                              final SortedSet<BinaryDomainCombination> all_bdcs,
 545                                                              final Map<String, Set<BinaryDomainCombination>> bdcs_per_genome )
 546             throws IOException {
 547         int counter = 0;
 548         int total_absense_counter = 0;
 549         int not_total_absense_counter = 0;
 550         SurfacingUtil.addHtmlHead( html_writer, "Binary Domain Combination Copy Differences" );
 551         html_writer.write( "<body><table>" );
 552         for( final BinaryDomainCombination bdc : all_bdcs ) {
 553             if ( ( high_copy_base_values.get( bdc ) > 0 ) && ( high_copy_target_values.get( bdc ) > 0 )
 554                     && ( high_copy_base_values.get( bdc ) >= low_copy_values.get( bdc ) ) ) {
 555                 if ( high_copy_target_values.get( bdc ) >= min_diff + ( factor * low_copy_values.get( bdc ) ) ) {
 556                     if ( low_copy_values.get( bdc ) <= 0.0 ) {
 557                         ++total_absense_counter;
 558                     }
 559                     else {
 560                         ++not_total_absense_counter;
 561                     }
 562                     ++counter;
 563                     html_writer.write( "<tr><td><a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK + bdc.getId0()
 564                             + "\">" + bdc.getId0() + "</a> = <a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK
 565                             + bdc.getId1() + "\">" + bdc.getId1() + "</a>" );
 566                     html_writer.write( "</td><td>" );
 567                     html_writer.write( "<table>" );
 568                     for( final GenomeWideCombinableDomains genome : genomes ) {
 569                         final String species = genome.getSpecies().getSpeciesId();
 570                         if ( high_copy_target_species.contains( species ) ) {
 571                             html_writer.write( "<tr>" );
 572                             writeCopyNumberValues( high_copy_target_values,
 573                                                    bdc,
 574                                                    genome,
 575                                                    bdcs_per_genome,
 576                                                    species,
 577                                                    html_writer,
 578                                                    "#0000FF" );
 579                             html_writer.write( "</tr>" );
 580                         }
 581                         else if ( low_copy_species.contains( species ) ) {
 582                             html_writer.write( "<tr>" );
 583                             writeCopyNumberValues( low_copy_values,
 584                                                    bdc,
 585                                                    genome,
 586                                                    bdcs_per_genome,
 587                                                    species,
 588                                                    html_writer,
 589                                                    "#A0A0A0" );
 590                             html_writer.write( "</tr>" );
 591                         }
 592                         else if ( high_copy_base_species.contains( species ) ) {
 593                             html_writer.write( "<tr>" );
 594                             writeCopyNumberValues( high_copy_base_values,
 595                                                    bdc,
 596                                                    genome,
 597                                                    bdcs_per_genome,
 598                                                    species,
 599                                                    html_writer,
 600                                                    "#404040" );
 601                             html_writer.write( "</tr>" );
 602                         }
 603                     }
 604                     html_writer.write( "</table>" );
 605                     html_writer.write( "</td></tr>" );
 606                     html_writer.write( SurfacingConstants.NL );
 607                 }
 608             }
 609         }
 610         html_writer.write( "</table>" );
 611         html_writer.write( SurfacingConstants.NL );
 612         html_writer.write( "<hr>" );
 613         html_writer.write( SurfacingConstants.NL );
 614         html_writer.write( "Rule 1: high-copy-base > 0 && high-copy-target > 0 && high-copy-base >= low-copy" );
 615         html_writer.write( "<br>" );
 616         html_writer.write( SurfacingConstants.NL );
 617         html_writer.write( "Rule 2: high-copy-target >= minimal-difference + ( factor * low-copy )" );
 618         html_writer.write( "<br>" );
 619         html_writer.write( SurfacingConstants.NL );
 620         html_writer.write( "Calculation mode for high copy target : " + COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES );
 621         html_writer.write( SurfacingConstants.NL );
 622         html_writer.write( "<br>" );
 623         html_writer.write( "Calculation mode for high copy base : " + COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES );
 624         html_writer.write( SurfacingConstants.NL );
 625         html_writer.write( "<br>" );
 626         html_writer.write( "Calculation mode for low copy : " + COPY_CALC_MODE_FOR_LOW_COPY_SPECIES );
 627         html_writer.write( SurfacingConstants.NL );
 628         html_writer.write( "<br>" );
 629         html_writer.write( "Minimal difference : " + min_diff );
 630         html_writer.write( SurfacingConstants.NL );
 631         html_writer.write( "<br>" );
 632         html_writer.write( "Factor : " + factor );
 633         html_writer.write( SurfacingConstants.NL );
 634         html_writer.write( "<br>" );
 635         html_writer.write( "Lower copy binary domain combinations : " + counter );
 636         html_writer.write( SurfacingConstants.NL );
 637         html_writer.write( "<br>" );
 638         html_writer.write( "Total absence : " + total_absense_counter );
 639         html_writer.write( SurfacingConstants.NL );
 640         html_writer.write( "<br>" );
 641         html_writer.write( "Not total absence : " + not_total_absense_counter );
 642         html_writer.write( SurfacingConstants.NL );
 643         html_writer.write( "<br>" );
 644         html_writer.write( "Total binary domain combinations : " + all_bdcs.size() );
 645         html_writer.write( SurfacingConstants.NL );
 646         html_writer.write( "<hr>" );
 647         html_writer.write( SurfacingConstants.NL );
 648         html_writer.write( "</body></html>" );
 649         html_writer.write( SurfacingConstants.NL );
 650         html_writer.close();
 651     }
 652
 653     private static void writeDomainValuesToFiles( final List<GenomeWideCombinableDomains> genomes,
 654                                                   final List<String> high_copy_base_species,
 655                                                   final List<String> high_copy_target_species,
 656                                                   final List<String> low_copy_species,
 657                                                   final int min_diff,
 658                                                   final Double factor,
 659                                                   final Map<DomainId, List<GoId>> domain_id_to_go_ids_map,
 660                                                   final Map<GoId, GoTerm> go_id_to_term_map,
 661                                                   final Writer plain_writer,
 662                                                   final Writer html_writer,
 663                                                   final File proteins_file_base,
 664                                                   final SortedMap<DomainId, Double> high_copy_base_values,
 665                                                   final SortedMap<DomainId, Double> high_copy_target_values,
 666                                                   final SortedMap<DomainId, Double> low_copy_values,
 667                                                   final SortedSet<DomainId> all_domains,
 668                                                   final SortedSet<GoId> go_ids_of_passing_domains,
 669                                                   final SortedMap<Species, List<Protein>> protein_lists_per_species )
 670             throws IOException {
 671         int counter = 0;
 672         int total_absense_counter = 0;
 673         int not_total_absense_counter = 0;
 674         SurfacingUtil.addHtmlHead( html_writer, "Domain Copy Differences" );
 675         html_writer.write( "<body><table>" );
 676         for( final DomainId domain_id : all_domains ) {
 677             if ( ( high_copy_base_values.get( domain_id ) > 0 ) && ( high_copy_target_values.get( domain_id ) > 0 )
 678                     && ( high_copy_base_values.get( domain_id ) >= low_copy_values.get( domain_id ) ) ) {
 679                 if ( high_copy_target_values.get( domain_id ) >= min_diff
 680                         + ( factor * low_copy_values.get( domain_id ) ) ) {
 681                     if ( low_copy_values.get( domain_id ) <= 0.0 ) {
 682                         ++total_absense_counter;
 683                     }
 684                     else {
 685                         ++not_total_absense_counter;
 686                     }
 687                     ++counter;
 688                     writeProteinsToFile( proteins_file_base, protein_lists_per_species, domain_id );
 689                     if ( domain_id_to_go_ids_map.containsKey( domain_id ) ) {
 690                         go_ids_of_passing_domains.addAll( domain_id_to_go_ids_map.get( domain_id ) );
 691                     }
 692                     plain_writer.write( domain_id.getId() );
 693                     plain_writer.write( SurfacingConstants.NL );
 694                     html_writer.write( "<tr><td><a href=\"" + SurfacingConstants.PFAM_FAMILY_ID_LINK
 695                             + domain_id.getId() + "\">" + domain_id.getId() + "</a></td><td>" );
 696                     html_writer.write( addGoInformation( domain_id, domain_id_to_go_ids_map, go_id_to_term_map )
 697                             .toString() );
 698                     html_writer.write( "</td><td>" );
 699                     html_writer.write( "<table>" );
 700                     for( final GenomeWideCombinableDomains genome : genomes ) {
 701                         final String species = genome.getSpecies().getSpeciesId();
 702                         if ( high_copy_target_species.contains( species ) ) {
 703                             html_writer.write( "<tr>" );
 704                             writeCopyNumberValues( high_copy_target_values,
 705                                                    domain_id,
 706                                                    genome,
 707                                                    species,
 708                                                    plain_writer,
 709                                                    html_writer,
 710                                                    "#0000FF" );
 711                             html_writer.write( "</tr>" );
 712                         }
 713                         else if ( low_copy_species.contains( species ) ) {
 714                             html_writer.write( "<tr>" );
 715                             writeCopyNumberValues( low_copy_values,
 716                                                    domain_id,
 717                                                    genome,
 718                                                    species,
 719                                                    plain_writer,
 720                                                    html_writer,
 721                                                    "#A0A0A0" );
 722                             html_writer.write( "</tr>" );
 723                         }
 724                         else if ( high_copy_base_species.contains( species ) ) {
 725                             html_writer.write( "<tr>" );
 726                             writeCopyNumberValues( high_copy_base_values,
 727                                                    domain_id,
 728                                                    genome,
 729                                                    species,
 730                                                    plain_writer,
 731                                                    html_writer,
 732                                                    "#404040" );
 733                             html_writer.write( "</tr>" );
 734                         }
 735                     }
 736                     html_writer.write( "</table>" );
 737                     html_writer.write( "</td></tr>" );
 738                     html_writer.write( SurfacingConstants.NL );
 739                     plain_writer.write( SurfacingConstants.NL );
 740                 }
 741             }
 742         }
 743         html_writer.write( "</table>" );
 744         html_writer.write( SurfacingConstants.NL );
 745         html_writer.write( "<hr>" );
 746         html_writer.write( SurfacingConstants.NL );
 747         html_writer.write( "Rule 1: high-copy-base > 0 && high-copy-target > 0 && high-copy-base >= low-copy" );
 748         html_writer.write( "<br>" );
 749         html_writer.write( SurfacingConstants.NL );
 750         html_writer.write( "Rule 2: high-copy-target >= minimal-difference + ( factor * low-copy )" );
 751         html_writer.write( "<br>" );
 752         html_writer.write( SurfacingConstants.NL );
 753         html_writer.write( "Calculation mode for high copy target : " + COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES );
 754         html_writer.write( SurfacingConstants.NL );
 755         html_writer.write( "<br>" );
 756         html_writer.write( "Calculation mode for high copy base : " + COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES );
 757         html_writer.write( SurfacingConstants.NL );
 758         html_writer.write( "<br>" );
 759         html_writer.write( "Calculation mode for low copy : " + COPY_CALC_MODE_FOR_LOW_COPY_SPECIES );
 760         html_writer.write( SurfacingConstants.NL );
 761         html_writer.write( "<br>" );
 762         html_writer.write( "Minimal difference : " + min_diff );
 763         html_writer.write( SurfacingConstants.NL );
 764         html_writer.write( "<br>" );
 765         html_writer.write( "Factor : " + factor );
 766         html_writer.write( SurfacingConstants.NL );
 767         html_writer.write( "<br>" );
 768         html_writer.write( "Lower copy domains : " + counter );
 769         html_writer.write( SurfacingConstants.NL );
 770         html_writer.write( "<br>" );
 771         html_writer.write( "Total absence : " + total_absense_counter );
 772         html_writer.write( SurfacingConstants.NL );
 773         html_writer.write( "<br>" );
 774         html_writer.write( "Not total absence : " + not_total_absense_counter );
 775         html_writer.write( SurfacingConstants.NL );
 776         html_writer.write( "<br>" );
 777         html_writer.write( "Total domains : " + all_domains.size() );
 778         html_writer.write( SurfacingConstants.NL );
 779         html_writer.write( "<hr>" );
 780         html_writer.write( SurfacingConstants.NL );
 781         html_writer.write( "</body></html>" );
 782         html_writer.write( SurfacingConstants.NL );
 783         html_writer.close();
 784         plain_writer.write( "# Rule 1: high-copy-base > 0 && high-copy-target > 0 && high-copy-base >= low-copy" );
 785         plain_writer.write( SurfacingConstants.NL );
 786         plain_writer.write( "# Rule 2: high-copy-target >= minimal-difference + ( factor * low-copy )" );
 787         plain_writer.write( SurfacingConstants.NL );
 788         plain_writer.write( "# Calculation mode for high copy target: " + COPY_CALC_MODE_FOR_HIGH_COPY_TARGET_SPECIES );
 789         plain_writer.write( SurfacingConstants.NL );
 790         plain_writer.write( "# Calculation mode for high copy base  : " + COPY_CALC_MODE_FOR_HIGH_COPY_BASE_SPECIES );
 791         plain_writer.write( SurfacingConstants.NL );
 792         plain_writer.write( "# Calculation mode for low copy        : " + COPY_CALC_MODE_FOR_LOW_COPY_SPECIES );
 793         plain_writer.write( SurfacingConstants.NL );
 794         plain_writer.write( "# Minimal difference: " + min_diff );
 795         plain_writer.write( SurfacingConstants.NL );
 796         plain_writer.write( "# Factor            : " + factor );
 797         plain_writer.write( SurfacingConstants.NL );
 798         plain_writer.write( "# Lower copy domains: " + counter );
 799         plain_writer.write( SurfacingConstants.NL );
 800         plain_writer.write( "# Total absence     : " + total_absense_counter );
 801         plain_writer.write( SurfacingConstants.NL );
 802         plain_writer.write( "# Not total absence : " + not_total_absense_counter );
 803         plain_writer.write( SurfacingConstants.NL );
 804         plain_writer.write( "# Total domains     : " + all_domains.size() );
 805         plain_writer.write( SurfacingConstants.NL );
 806         plain_writer.close();
 807     }
 808
 809     private static void writeGoIdsToFile( final Writer writer, final SortedSet<GoId> gos ) throws IOException {
 810         for( final GoId go_id : gos ) {
 811             writer.write( go_id.toString() );
 812             writer.write( SurfacingConstants.NL );
 813         }
 814         writer.close();
 815     }
 816
 817     private static void writeProteinsToFile( final File proteins_file_base,
 818                                              final SortedMap<Species, List<Protein>> protein_lists_per_species,
 819                                              final DomainId domain_id ) throws IOException {
 820         final File my_proteins_file = new File( proteins_file_base.getParentFile() + ForesterUtil.FILE_SEPARATOR
 821                 + domain_id + PLUS_MINUS_PROTEINS_FILE_DOM_SUFFIX );
 822         SurfacingUtil.checkForOutputFileWriteability( my_proteins_file );
 823         final Writer proteins_file_writer = new BufferedWriter( new FileWriter( my_proteins_file ) );
 824         SurfacingUtil.extractProteinNames( protein_lists_per_species,
 825                                            domain_id,
 826                                            proteins_file_writer,
 827                                            "\t",
 828                                            surfacing.LIMIT_SPEC_FOR_PROT_EX );
 829         proteins_file_writer.close();
 830         System.out.println( "Wrote proteins list to \"" + my_proteins_file + "\"" );
 831     }
 832
 833     public static enum COPY_CALCULATION_MODE {
 834         MEAN, MEDIAN, MAX, MIN
 835     }
 836 }