forester/java/src/org/forester/application/gsdi.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2009 Christian M. Zmasek
   6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   7 // All rights reserved
   8 //
   9 // This library is free software; you can redistribute it and/or
  10 // modify it under the terms of the GNU Lesser General Public
  11 // License as published by the Free Software Foundation; either
  12 // version 2.1 of the License, or (at your option) any later version.
  13 //
  14 // This library is distributed in the hope that it will be useful,
  15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 // Lesser General Public License for more details.
  18 //
  19 // You should have received a copy of the GNU Lesser General Public
  20 // License along with this library; if not, write to the Free Software
  21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  22 //
  23 // Contact: phylosoft @ gmail . com
  24 // WWW: www.phylosoft.org/forester
  25
  26 package org.forester.application;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.text.SimpleDateFormat;
  31 import java.util.ArrayList;
  32 import java.util.Date;
  33 import java.util.List;
  34 import java.util.SortedMap;
  35 import java.util.SortedSet;
  36 import java.util.TreeMap;
  37 import java.util.TreeSet;
  38
  39 import org.forester.io.parsers.PhylogenyParser;
  40 import org.forester.io.parsers.nhx.NHXParser;
  41 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  42 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
  43 import org.forester.io.parsers.util.ParserUtils;
  44 import org.forester.io.writers.PhylogenyWriter;
  45 import org.forester.phylogeny.Phylogeny;
  46 import org.forester.phylogeny.PhylogenyMethods;
  47 import org.forester.phylogeny.PhylogenyNode;
  48 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  49 import org.forester.phylogeny.factories.PhylogenyFactory;
  50 import org.forester.sdi.GSDI;
  51 import org.forester.sdi.GSDIR;
  52 import org.forester.sdi.SDI;
  53 import org.forester.sdi.SDI.TaxonomyComparisonBase;
  54 import org.forester.sdi.SDIException;
  55 import org.forester.sdi.SDIse;
  56 import org.forester.util.CommandLineArguments;
  57 import org.forester.util.EasyWriter;
  58 import org.forester.util.ForesterConstants;
  59 import org.forester.util.ForesterUtil;
  60
  61 public final class gsdi {
  62
  63     private enum BASE_ALGORITHM {
  64         GSDIR, GSDI, SDI
  65     }
  66     final static public boolean REPLACE_UNDERSCORES_IN_NH_SPECIES_TREE = true;
  67     final static private String ALLOW_STRIPPING_OF_GENE_TREE_OPTION    = "g";
  68     final static private String SDISE_OPTION                           = "b";
  69     final static private String GSDIR_OPTION                           = "r";
  70     final static private String MOST_PARSIMONIOUS_OPTION               = "m";
  71     final static private String GUESS_FORMAT_OF_SPECIES_TREE           = "q";
  72     final static private String HELP_OPTION_1                          = "help";
  73     final static private String HELP_OPTION_2                          = "h";
  74     final static private String SUFFIX_FOR_SPECIES_TREE_USED           = "_species_tree_used.xml";
  75     final static private String LOGFILE_SUFFIX                         = "_gsdi_log.txt";
  76     final static private String REMAPPED_SUFFIX                        = "_gsdi_remapped.txt";
  77     final static private String PRG_NAME                               = "gsdi";
  78     final static private String PRG_VERSION                            = "1.000";
  79     final static private String PRG_DATE                               = "120629";
  80     final static private String PRG_DESC                               = "general speciation duplication inference";
  81     final static private String E_MAIL                                 = "phylosoft@gmail.com";
  82     final static private String WWW                                    = "www.phylosoft.org/forester";
  83
  84     public static void main( final String args[] ) {
  85         try {
  86             ForesterUtil.printProgramInformation( PRG_NAME,
  87                                                   PRG_DESC,
  88                                                   PRG_VERSION,
  89                                                   PRG_DATE,
  90                                                   E_MAIL,
  91                                                   WWW,
  92                                                   ForesterUtil.getForesterLibraryInformation() );
  93             CommandLineArguments cla = null;
  94             try {
  95                 cla = new CommandLineArguments( args );
  96             }
  97             catch ( final Exception e ) {
  98                 ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  99             }
 100             if ( cla.isOptionSet( gsdi.HELP_OPTION_1 ) || cla.isOptionSet( gsdi.HELP_OPTION_2 ) ) {
 101                 System.out.println();
 102                 gsdi.print_help();
 103                 System.exit( 0 );
 104             }
 105             else if ( ( args.length < 2 ) || ( cla.getNumberOfNames() != 3 ) ) {
 106                 System.out.println();
 107                 System.out.println( "Wrong number of arguments." );
 108                 System.out.println();
 109                 gsdi.print_help();
 110                 System.exit( -1 );
 111             }
 112             final List<String> allowed_options = new ArrayList<String>();
 113             allowed_options.add( gsdi.SDISE_OPTION );
 114             allowed_options.add( gsdi.GSDIR_OPTION );
 115             allowed_options.add( gsdi.GUESS_FORMAT_OF_SPECIES_TREE );
 116             allowed_options.add( gsdi.MOST_PARSIMONIOUS_OPTION );
 117             allowed_options.add( gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION );
 118             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 119             if ( dissallowed_options.length() > 0 ) {
 120                 ForesterUtil.fatalError( gsdi.PRG_NAME, "unknown option(s): " + dissallowed_options );
 121             }
 122             execute( cla );
 123         }
 124         catch ( final IOException e ) {
 125             ForesterUtil.fatalError( gsdi.PRG_NAME, e.getMessage() );
 126         }
 127     }
 128
 129     private static void execute( final CommandLineArguments cla ) throws IOException {
 130         BASE_ALGORITHM base_algorithm = BASE_ALGORITHM.GSDI;
 131         boolean most_parsimonous_duplication_model = false;
 132         boolean allow_stripping_of_gene_tree = false;
 133         if ( cla.isOptionSet( gsdi.GSDIR_OPTION ) ) {
 134             base_algorithm = BASE_ALGORITHM.GSDIR;
 135         }
 136         else if ( cla.isOptionSet( gsdi.SDISE_OPTION ) ) {
 137             base_algorithm = BASE_ALGORITHM.SDI;
 138         }
 139         if ( cla.isOptionSet( gsdi.MOST_PARSIMONIOUS_OPTION ) ) {
 140             if ( base_algorithm == BASE_ALGORITHM.SDI ) {
 141                 ForesterUtil.fatalError( gsdi.PRG_NAME, "Cannot use most parsimonious duplication mode with SDI" );
 142             }
 143             most_parsimonous_duplication_model = true;
 144         }
 145         if ( cla.isOptionSet( gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION ) ) {
 146             if ( base_algorithm == BASE_ALGORITHM.SDI ) {
 147                 ForesterUtil.fatalError( gsdi.PRG_NAME, "Cannot allow stripping of gene tree with SDI" );
 148             }
 149             allow_stripping_of_gene_tree = true;
 150         }
 151         Phylogeny species_tree = null;
 152         Phylogeny gene_tree = null;
 153         File gene_tree_file = null;
 154         File species_tree_file = null;
 155         File out_file = null;
 156         File log_file = null;
 157         EasyWriter log_writer = null;
 158         try {
 159             gene_tree_file = cla.getFile( 0 );
 160             species_tree_file = cla.getFile( 1 );
 161             out_file = cla.getFile( 2 );
 162             log_file = new File( ForesterUtil.removeSuffix( out_file.toString() ) + LOGFILE_SUFFIX );
 163         }
 164         catch ( final IllegalArgumentException e ) {
 165             ForesterUtil.fatalError( gsdi.PRG_NAME, "error in command line: " + e.getMessage() );
 166         }
 167         if ( ForesterUtil.isReadableFile( gene_tree_file ) != null ) {
 168             ForesterUtil.fatalError( gsdi.PRG_NAME, ForesterUtil.isReadableFile( gene_tree_file ) );
 169         }
 170         if ( ForesterUtil.isReadableFile( species_tree_file ) != null ) {
 171             ForesterUtil.fatalError( gsdi.PRG_NAME, ForesterUtil.isReadableFile( species_tree_file ) );
 172         }
 173         if ( ForesterUtil.isWritableFile( out_file ) != null ) {
 174             ForesterUtil.fatalError( gsdi.PRG_NAME, ForesterUtil.isWritableFile( out_file ) );
 175         }
 176         if ( ForesterUtil.isWritableFile( log_file ) != null ) {
 177             ForesterUtil.fatalError( gsdi.PRG_NAME, ForesterUtil.isWritableFile( log_file ) );
 178         }
 179         try {
 180             log_writer = ForesterUtil.createEasyWriter( log_file );
 181         }
 182         catch ( final IOException e ) {
 183             ForesterUtil.fatalError( gsdi.PRG_NAME, "Failed to create [" + log_file + "]: " + e.getMessage() );
 184         }
 185         try {
 186             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 187             gene_tree = factory.create( gene_tree_file, new PhyloXmlParser() )[ 0 ];
 188         }
 189         catch ( final IOException e ) {
 190             fatalError( "ERROR",
 191                         "Failed to read gene tree from [" + gene_tree_file + "]: " + e.getMessage(),
 192                         log_writer );
 193         }
 194         try {
 195             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 196             final PhylogenyParser p = ParserUtils.createParserDependingOnFileType( species_tree_file, true );
 197             if ( p instanceof PhyloXmlParser ) {
 198                 species_tree = factory.create( species_tree_file, p )[ 0 ];
 199             }
 200             else {
 201                 if ( REPLACE_UNDERSCORES_IN_NH_SPECIES_TREE && ( p instanceof NHXParser ) ) {
 202                     ( ( NHXParser ) p ).setReplaceUnderscores( true );
 203                 }
 204                 species_tree = factory.create( species_tree_file, p )[ 0 ];
 205                 final TaxonomyComparisonBase comp_base = GSDI.determineTaxonomyComparisonBase( gene_tree );
 206                 switch ( comp_base ) {
 207                     case SCIENTIFIC_NAME:
 208                         try {
 209                             PhylogenyMethods
 210                                     .transferNodeNameToField( species_tree,
 211                                                               PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
 212                                                               true );
 213                         }
 214                         catch ( final PhyloXmlDataFormatException e ) {
 215                             fatalError( "USER ERROR", "Failed to transfer general node name to scientific name, in ["
 216                                     + species_tree_file + "]: " + e.getMessage(), log_writer );
 217                         }
 218                         break;
 219                     case CODE:
 220                         try {
 221                             PhylogenyMethods
 222                                     .transferNodeNameToField( species_tree,
 223                                                               PhylogenyMethods.PhylogenyNodeField.TAXONOMY_CODE,
 224                                                               true );
 225                         }
 226                         catch ( final PhyloXmlDataFormatException e ) {
 227                             fatalError( "USER ERROR", "Failed to transfer general node name to taxonomy code, in ["
 228                                     + species_tree_file + "]: " + e.getMessage(), log_writer );
 229                         }
 230                         break;
 231                     case ID:
 232                         try {
 233                             PhylogenyMethods.transferNodeNameToField( species_tree,
 234                                                                       PhylogenyMethods.PhylogenyNodeField.TAXONOMY_ID,
 235                                                                       true );
 236                         }
 237                         catch ( final PhyloXmlDataFormatException e ) {
 238                             fatalError( "USER ERROR", "Failed to transfer general node name to taxonomy id, in ["
 239                                     + species_tree_file + "]: " + e.getMessage(), log_writer );
 240                         }
 241                         break;
 242                     default:
 243                         fatalError( "UNEXPECTED ERROR", "unable to determine comparison base", log_writer );
 244                 }
 245             }
 246         }
 247         catch ( final IOException e ) {
 248             fatalError( "ERROR",
 249                         "Failed to read species tree from [" + species_tree_file + "]: " + e.getMessage(),
 250                         log_writer );
 251         }
 252         gene_tree.setRooted( true );
 253         species_tree.setRooted( true );
 254         if ( !gene_tree.isCompletelyBinary() ) {
 255             fatalError( "user error", "gene tree is not completely binary", log_writer );
 256         }
 257         if ( base_algorithm == BASE_ALGORITHM.SDI ) {
 258             if ( !species_tree.isCompletelyBinary() ) {
 259                 fatalError( "user error",
 260                             "species tree is not completely binary, use GSDI or GSDIR instead",
 261                             log_writer );
 262             }
 263         }
 264         log_writer.println( PRG_NAME + " - " + PRG_DESC );
 265         log_writer.println( "  version         : " + PRG_VERSION );
 266         log_writer.println( "  date            : " + PRG_DATE );
 267         log_writer.println( "  forester version: " + ForesterConstants.FORESTER_VERSION );
 268         log_writer.println();
 269         log_writer.println( "Start time                               : "
 270                 + new SimpleDateFormat( "yyyyMMdd HH:mm:ss" ).format( new Date() ) );
 271         System.out.println( "Start time                               : "
 272                 + new SimpleDateFormat( "yyyyMMdd HH:mm:ss" ).format( new Date() ) );
 273         log_writer.println( "Gene tree file                           : " + gene_tree_file.getCanonicalPath() );
 274         System.out.println( "Gene tree file                           : " + gene_tree_file.getCanonicalPath() );
 275         log_writer.println( "Gene tree name                           : "
 276                 + ( ForesterUtil.isEmpty( gene_tree.getName() ) ? "" : gene_tree.getName() ) );
 277         System.out.println( "Gene tree name                           : "
 278                 + ( ForesterUtil.isEmpty( gene_tree.getName() ) ? "" : gene_tree.getName() ) );
 279         log_writer.println( "Species tree file                        : " + species_tree_file.getCanonicalPath() );
 280         System.out.println( "Species tree file                        : " + species_tree_file.getCanonicalPath() );
 281         log_writer.println( "Species tree name                        : "
 282                 + ( ForesterUtil.isEmpty( species_tree.getName() ) ? "" : gene_tree.getName() ) );
 283         System.out.println( "Species tree name                        : "
 284                 + ( ForesterUtil.isEmpty( species_tree.getName() ) ? "" : gene_tree.getName() ) );
 285         SDI sdi = null;
 286         final long start_time = new Date().getTime();
 287         try {
 288             if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) || ( base_algorithm == BASE_ALGORITHM.GSDIR ) ) {
 289                 if ( base_algorithm == BASE_ALGORITHM.GSDI ) {
 290                     System.out.println( "Algorithm       : GSDI" );
 291                     log_writer.println( "Algorithm       : GSDI" );
 292                 }
 293                 else if ( base_algorithm == BASE_ALGORITHM.GSDIR ) {
 294                     System.out.println( "Algorithm       : GSDIR" );
 295                     log_writer.println( "Algorithm       : GSDIR" );
 296                 }
 297                 System.out.println( "Use most parsimonous duplication model   : " + most_parsimonous_duplication_model );
 298                 System.out.println( "Allow stripping of gene tree nodes       : " + allow_stripping_of_gene_tree );
 299                 log_writer.println( "Use most parsimonous duplication model   : " + most_parsimonous_duplication_model );
 300                 log_writer.println( "Allow stripping of gene tree nodes       : " + allow_stripping_of_gene_tree );
 301                 log_writer.flush();
 302                 if ( base_algorithm == BASE_ALGORITHM.GSDI ) {
 303                     sdi = new GSDI( gene_tree,
 304                                     species_tree,
 305                                     most_parsimonous_duplication_model,
 306                                     allow_stripping_of_gene_tree,
 307                                     true );
 308                 }
 309                 else if ( base_algorithm == BASE_ALGORITHM.GSDIR ) {
 310                     sdi = new GSDIR( gene_tree, species_tree, allow_stripping_of_gene_tree, 1 );
 311                 }
 312             }
 313             else {
 314                 System.out.println();
 315                 System.out.println( "Algorithm       : SDI" );
 316                 log_writer.println( "Algorithm       : SDI" );
 317                 log_writer.flush();
 318                 sdi = new SDIse( gene_tree, species_tree );
 319             }
 320         }
 321         catch ( final SDIException e ) {
 322             fatalError( "user error", e.getLocalizedMessage(), log_writer );
 323         }
 324         catch ( final IOException e ) {
 325             fatalError( "error", e.toString(), log_writer );
 326         }
 327         catch ( final Exception e ) {
 328             e.printStackTrace();
 329             fatalError( "unexpected error", e.toString(), log_writer );
 330         }
 331         System.out.println( "Running time (excluding I/O)             : " + ( new Date().getTime() - start_time )
 332                 + "ms" );
 333         log_writer.println( "Running time (excluding I/O)             : " + ( new Date().getTime() - start_time )
 334                 + "ms" );
 335         if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) ) {
 336             final GSDI gsdi = ( GSDI ) sdi;
 337             System.out.println( "Mapping based on                         : " + gsdi.getTaxCompBase() );
 338             log_writer.println( "Mapping based on                         : " + gsdi.getTaxCompBase() );
 339         }
 340         if ( ( base_algorithm == BASE_ALGORITHM.GSDIR ) ) {
 341             final GSDIR gsdir = ( GSDIR ) sdi;
 342             System.out.println( "Mapping based on                         : " + gsdir.getTaxCompBase() );
 343             log_writer.println( "Mapping based on                         : " + gsdir.getTaxCompBase() );
 344             System.out.println( "Minimal duplications sum                 : " + gsdir.getMinDuplicationsSum() );
 345             log_writer.println( "Minimal duplications sum                 : " + gsdir.getMinDuplicationsSum() );
 346             System.out.println( "Duplications sum statistics              : " + gsdir.getMinDuplicationsSum() );
 347             log_writer.println( "Duplications sum statistics              : " + gsdir.getMinDuplicationsSum() );
 348         }
 349         try {
 350             final PhylogenyWriter writer = new PhylogenyWriter();
 351             if ( base_algorithm == BASE_ALGORITHM.GSDIR ) {
 352                 writer.toPhyloXML( out_file,
 353                                    ( ( GSDIR ) sdi ).getMinDuplicationsSumGeneTrees(),
 354                                    0,
 355                                    ForesterUtil.LINE_SEPARATOR );
 356             }
 357             else {
 358                 writer.toPhyloXML( out_file, gene_tree, 0 );
 359             }
 360         }
 361         catch ( final IOException e ) {
 362             ForesterUtil.fatalError( PRG_NAME,
 363                                      "Failed to write to [" + out_file.getCanonicalPath() + "]: " + e.getMessage() );
 364         }
 365         System.out.println( "Wrote resulting gene tree to             : " + out_file.getCanonicalPath() );
 366         log_writer.println( "Wrote resulting gene tree to             : " + out_file.getCanonicalPath() );
 367         if ( base_algorithm == BASE_ALGORITHM.SDI ) {
 368             sdi.computeMappingCostL();
 369             System.out.println( "Mapping cost                             : " + sdi.computeMappingCostL() );
 370             log_writer.println( "Mapping cost                             : " + sdi.computeMappingCostL() );
 371         }
 372         else if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) || ( base_algorithm == BASE_ALGORITHM.GSDIR ) ) {
 373             final GSDI gsdi = ( GSDI ) sdi;
 374             final File species_tree_used_file = new File( ForesterUtil.removeSuffix( out_file.toString() )
 375                     + SUFFIX_FOR_SPECIES_TREE_USED );
 376             try {
 377                 final PhylogenyWriter writer = new PhylogenyWriter();
 378                 writer.toPhyloXML( species_tree_used_file, gsdi.getSpeciesTree(), 0 );
 379             }
 380             catch ( final IOException e ) {
 381                 ForesterUtil.fatalError( PRG_NAME, "Failed to write to [" + species_tree_used_file.getCanonicalPath()
 382                         + "]: " + e.getMessage() );
 383             }
 384             System.out.println( "Wrote (stripped) species tree to         : "
 385                     + species_tree_used_file.getCanonicalPath() );
 386             log_writer.println( "Wrote (stripped) species tree to         : "
 387                     + species_tree_used_file.getCanonicalPath() );
 388             if ( ( gsdi.getReMappedScientificNamesFromGeneTree() != null )
 389                     && !gsdi.getReMappedScientificNamesFromGeneTree().isEmpty() ) {
 390                 System.out.println( "Number of gene tree species remapped     : "
 391                         + gsdi.getReMappedScientificNamesFromGeneTree().size() );
 392                 log_writer.println( "Number of gene tree species remapped     : "
 393                         + gsdi.getReMappedScientificNamesFromGeneTree().size() );
 394                 writeToRemappedFile( out_file, gsdi.getReMappedScientificNamesFromGeneTree(), log_writer );
 395             }
 396         }
 397         System.out.println( "Number of external nodes in gene tree    : " + gene_tree.getNumberOfExternalNodes() );
 398         log_writer.println( "Number of external nodes in gene tree    : " + gene_tree.getNumberOfExternalNodes() );
 399         System.out.println( "Number of external nodes in species tree : "
 400                 + sdi.getSpeciesTree().getNumberOfExternalNodes() );
 401         log_writer.println( "Number of external nodes in species tree : "
 402                 + sdi.getSpeciesTree().getNumberOfExternalNodes() );
 403         if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) ) {
 404             final GSDI gsdi = ( GSDI ) sdi;
 405             final int poly = PhylogenyMethods.countNumberOfPolytomies( gsdi.getSpeciesTree() );
 406             System.out.println( "Number of polytomies in species tree     : " + poly );
 407             log_writer.println( "Number of polytomies in species tree     : " + poly );
 408             System.out.println( "External nodes stripped from gene tree   : "
 409                     + gsdi.getStrippedExternalGeneTreeNodes().size() );
 410             log_writer.println( "External nodes stripped from gene tree   : "
 411                     + gsdi.getStrippedExternalGeneTreeNodes().size() );
 412             System.out.println( "External nodes stripped from species tree: "
 413                     + gsdi.getStrippedSpeciesTreeNodes().size() );
 414             log_writer.println( "External nodes stripped from species tree: "
 415                     + gsdi.getStrippedSpeciesTreeNodes().size() );
 416         }
 417         System.out.println();
 418         System.out.println( "Number of duplications                   : " + sdi.getDuplicationsSum() );
 419         log_writer.println( "Number of duplications                   : " + sdi.getDuplicationsSum() );
 420         if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) ) {
 421             final GSDI gsdi = ( GSDI ) sdi;
 422             if ( !most_parsimonous_duplication_model ) {
 423                 final int u = gsdi.getSpeciationOrDuplicationEventsSum();
 424                 System.out.println( "Number of potential duplications         : " + u );
 425                 log_writer.println( "Number of potential duplications         : " + u );
 426             }
 427             System.out.println( "Number of speciations                    : " + gsdi.getSpeciationsSum() );
 428             log_writer.println( "Number of speciations                    : " + gsdi.getSpeciationsSum() );
 429             log_writer.println();
 430             printMappedNodesToLog( log_writer, gsdi );
 431             log_writer.println();
 432             printStrippedGeneTreeNodesToLog( log_writer, gsdi );
 433         }
 434         System.out.println();
 435         System.out.println( "Wrote log to                             : " + log_file.getCanonicalPath() );
 436         System.out.println();
 437         log_writer.close();
 438     }
 439
 440     private static void writeToRemappedFile( final File out_file,
 441                                              final SortedSet<String> remapped,
 442                                              final EasyWriter log_writer ) throws IOException {
 443         final File file = new File( ForesterUtil.removeSuffix( out_file.toString() ) + REMAPPED_SUFFIX );
 444         final EasyWriter remapped_writer = ForesterUtil.createEasyWriter( file );
 445         for( final String s : remapped ) {
 446             remapped_writer.println( s );
 447         }
 448         remapped_writer.close();
 449         System.out.println( "Wrote remapped gene tree species to      : " + file.getCanonicalPath() );
 450         log_writer.println( "Wrote remapped gene tree species to      : " + file.getCanonicalPath() );
 451     }
 452
 453     private static void printMappedNodesToLog( final EasyWriter log_writer, final GSDI gsdi ) throws IOException {
 454         final SortedSet<String> ss = new TreeSet<String>();
 455         for( final PhylogenyNode n : gsdi.getMappedExternalSpeciesTreeNodes() ) {
 456             ss.add( n.toString() );
 457         }
 458         log_writer.println( "The following " + ss.size() + " species were used: " );
 459         for( final String s : ss ) {
 460             log_writer.println( "  " + s );
 461         }
 462     }
 463
 464     private static void fatalError( final String type, final String msg, final EasyWriter log_writer ) {
 465         try {
 466             log_writer.flush();
 467             log_writer.println();
 468             log_writer.print( type.toUpperCase() + ": " );
 469             log_writer.println( msg );
 470             log_writer.close();
 471         }
 472         catch ( final IOException e ) {
 473             e.printStackTrace();
 474         }
 475         ForesterUtil.fatalError( gsdi.PRG_NAME, msg );
 476     }
 477
 478     private static void printStrippedGeneTreeNodesToLog( final EasyWriter log_writer, final GSDI gsdi )
 479             throws IOException {
 480         final SortedMap<String, Integer> sm = new TreeMap<String, Integer>();
 481         for( final PhylogenyNode n : gsdi.getStrippedExternalGeneTreeNodes() ) {
 482             final String s = n.toString();
 483             if ( sm.containsKey( s ) ) {
 484                 sm.put( s, sm.get( s ) + 1 );
 485             }
 486             else {
 487                 sm.put( s, 1 );
 488             }
 489         }
 490         log_writer.println( "The following " + sm.size() + " nodes were stripped from the gene tree: " );
 491         for( final String s : sm.keySet() ) {
 492             final int count = sm.get( s );
 493             if ( count == 1 ) {
 494                 log_writer.println( "  " + s );
 495             }
 496             else {
 497                 log_writer.println( "  " + s + " [" + count + "]" );
 498             }
 499         }
 500     }
 501
 502     private static void print_help() {
 503         System.out.println( "Usage: " + gsdi.PRG_NAME
 504                 + " [-options] <gene tree in phyloXML format> <species tree> <outfile>" );
 505         System.out.println();
 506         System.out.println( "Options:" );
 507         System.out.println( " -" + gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION
 508                 + ": to allow stripping of gene tree nodes without a matching species" );
 509         System.out.println( " -" + gsdi.MOST_PARSIMONIOUS_OPTION
 510                 + ": use most parimonious duplication model for GSDI: " );
 511         System.out.println( "     assign nodes as speciations which would otherwise be assiged" );
 512         System.out.println( "     as potential duplications due to polytomies in the species tree" );
 513         System.out.println( " -" + gsdi.GUESS_FORMAT_OF_SPECIES_TREE
 514                 + ": to allow species tree in other formats than phyloXML (i.e. Newick, NHX, Nexus)" );
 515         System.out.println( " -" + gsdi.SDISE_OPTION
 516                 + ": to use SDIse algorithm instead of GSDI algorithm (for binary species trees)" );
 517         System.out.println( " -" + gsdi.GSDIR_OPTION
 518                 + ": to use GSDIR algorithm instead of GSDI algorithm (re-rooting)" );
 519         System.out.println();
 520         System.out.println( "Gene tree:" );
 521         System.out.println( " in phyloXM format, with taxonomy and sequence data in appropriate fields" );
 522         System.out.println();
 523         System.out.println( "Species tree:" );
 524         System.out.println( " in phyloXML format (unless option -" + gsdi.GUESS_FORMAT_OF_SPECIES_TREE + " is used)" );
 525         System.out.println();
 526         System.out.println( "Example: gsdi -" + ALLOW_STRIPPING_OF_GENE_TREE_OPTION
 527                 + " gene_tree.xml tree_of_life.xml out.xml" );
 528         System.out.println();
 529     }
 530 }