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
  35 import org.forester.io.parsers.PhylogenyParser;
  36 import org.forester.io.parsers.nhx.NHXParser;
  37 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  38 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
  39 import org.forester.io.parsers.util.ParserUtils;
  40 import org.forester.io.writers.PhylogenyWriter;
  41 import org.forester.phylogeny.Phylogeny;
  42 import org.forester.phylogeny.PhylogenyMethods;
  43 import org.forester.phylogeny.PhylogenyNode;
  44 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  45 import org.forester.phylogeny.factories.PhylogenyFactory;
  46 import org.forester.sdi.GSDI;
  47 import org.forester.sdi.SDI;
  48 import org.forester.sdi.SDI.TaxonomyComparisonBase;
  49 import org.forester.sdi.SDIse;
  50 import org.forester.sdi.SdiException;
  51 import org.forester.util.CommandLineArguments;
  52 import org.forester.util.EasyWriter;
  53 import org.forester.util.ForesterConstants;
  54 import org.forester.util.ForesterUtil;
  55
  56 public final class gsdi {
  57
  58     private enum BASE_ALGORITHM {
  59         GSDI, SDI
  60     }
  61     final static public boolean REPLACE_UNDERSCORES_IN_NH_SPECIES_TREE     = true;
  62     final static private String ALLOW_STRIPPING_OF_GENE_TREE_OPTION        = "g";
  63     final static private String SDI_OPTION                                 = "b";
  64     final static private String MOST_PARSIMONIOUS_OPTION                   = "m";
  65     final static private String GUESS_FORMAT_OF_SPECIES_TREE               = "q";
  66     final static private String HELP_OPTION_1                              = "help";
  67     final static private String HELP_OPTION_2                              = "h";
  68     final static private String DEFAULT_OUTFILE_SUFFIX                     = "_gsdi_out.xml";
  69     final static private String SUFFIX_FOR_LIST_OF_STIPPED_GENE_TREE_NODES = "_stripped_gene_tree_nodes.txt";
  70     final static private String SUFFIX_FOR_SPECIES_TREE_USED               = "_species_tree_used.xml";
  71     final static private String LOGFILE_SUFFIX                             = "_gsdi_log.txt";
  72     final static private String PRG_NAME                                   = "gsdi";
  73     final static private String PRG_VERSION                                = "0.901";
  74     final static private String PRG_DATE                                   = "120608";
  75     final static private String PRG_DESC                                   = "general speciation duplication inference";
  76     final static private String E_MAIL                                     = "phylosoft@gmail.com";
  77     final static private String WWW                                        = "www.phylosoft.org/forester";
  78
  79     public static void main( final String args[] ) {
  80         try {
  81             ForesterUtil.printProgramInformation( PRG_NAME,
  82                                                   PRG_DESC,
  83                                                   PRG_VERSION,
  84                                                   PRG_DATE,
  85                                                   E_MAIL,
  86                                                   WWW,
  87                                                   ForesterUtil.getForesterLibraryInformation() );
  88             CommandLineArguments cla = null;
  89             try {
  90                 cla = new CommandLineArguments( args );
  91             }
  92             catch ( final Exception e ) {
  93                 ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  94             }
  95             if ( cla.isOptionSet( gsdi.HELP_OPTION_1 ) || cla.isOptionSet( gsdi.HELP_OPTION_2 ) ) {
  96                 System.out.println();
  97                 gsdi.print_help();
  98                 System.exit( 0 );
  99             }
 100             else if ( ( args.length < 2 ) || ( cla.getNumberOfNames() < 2 ) || ( cla.getNumberOfNames() > 3 ) ) {
 101                 System.out.println();
 102                 System.out.println( "Wrong number of arguments." );
 103                 System.out.println();
 104                 gsdi.print_help();
 105                 System.exit( -1 );
 106             }
 107             final List<String> allowed_options = new ArrayList<String>();
 108             allowed_options.add( gsdi.SDI_OPTION );
 109             allowed_options.add( gsdi.GUESS_FORMAT_OF_SPECIES_TREE );
 110             allowed_options.add( gsdi.MOST_PARSIMONIOUS_OPTION );
 111             allowed_options.add( gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION );
 112             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 113             if ( dissallowed_options.length() > 0 ) {
 114                 ForesterUtil.fatalError( gsdi.PRG_NAME, "unknown option(s): " + dissallowed_options );
 115             }
 116             execute( cla );
 117         }
 118         catch ( final IOException e ) {
 119             ForesterUtil.fatalError( gsdi.PRG_NAME, e.getMessage() );
 120         }
 121     }
 122
 123     private static void execute( final CommandLineArguments cla ) throws IOException {
 124         BASE_ALGORITHM base_algorithm = BASE_ALGORITHM.GSDI;
 125         boolean most_parsimonous_duplication_model = false;
 126         boolean species_tree_in_phyloxml = true;
 127         boolean allow_stripping_of_gene_tree = false;
 128         if ( cla.isOptionSet( gsdi.SDI_OPTION ) ) {
 129             base_algorithm = BASE_ALGORITHM.SDI;
 130         }
 131         if ( cla.isOptionSet( gsdi.MOST_PARSIMONIOUS_OPTION ) ) {
 132             if ( base_algorithm != BASE_ALGORITHM.GSDI ) {
 133                 ForesterUtil.fatalError( gsdi.PRG_NAME, "Can only use most parsimonious duplication mode with GSDI" );
 134             }
 135             most_parsimonous_duplication_model = true;
 136         }
 137         if ( cla.isOptionSet( gsdi.GUESS_FORMAT_OF_SPECIES_TREE ) ) {
 138             species_tree_in_phyloxml = false;
 139         }
 140         if ( cla.isOptionSet( gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION ) ) {
 141             if ( base_algorithm != BASE_ALGORITHM.GSDI ) {
 142                 ForesterUtil.fatalError( gsdi.PRG_NAME, "Can only allow stripping of gene tree with GSDI" );
 143             }
 144             allow_stripping_of_gene_tree = true;
 145         }
 146         Phylogeny species_tree = null;
 147         Phylogeny gene_tree = null;
 148         File gene_tree_file = null;
 149         File species_tree_file = null;
 150         File out_file = null;
 151         File log_file = null;
 152         EasyWriter log_writer = null;
 153         try {
 154             gene_tree_file = cla.getFile( 0 );
 155             species_tree_file = cla.getFile( 1 );
 156             if ( cla.getNumberOfNames() == 3 ) {
 157                 out_file = cla.getFile( 2 );
 158             }
 159             else {
 160                 out_file = new File( ForesterUtil.removeSuffix( gene_tree_file.toString() ) + DEFAULT_OUTFILE_SUFFIX );
 161             }
 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             ForesterUtil.fatalError( gsdi.PRG_NAME,
 191                                      "Failed to read gene tree from [" + gene_tree_file + "]: " + e.getMessage() );
 192         }
 193         try {
 194             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 195             if ( species_tree_in_phyloxml ) {
 196                 species_tree = factory.create( species_tree_file, new PhyloXmlParser() )[ 0 ];
 197             }
 198             else {
 199                 final PhylogenyParser p = ParserUtils.createParserDependingOnFileType( species_tree_file, true );
 200                 if ( REPLACE_UNDERSCORES_IN_NH_SPECIES_TREE && ( p instanceof NHXParser ) ) {
 201                     ( ( NHXParser ) p ).setReplaceUnderscores( true );
 202                 }
 203                 species_tree = factory.create( species_tree_file, p )[ 0 ];
 204                 final TaxonomyComparisonBase comp_base = GSDI.determineTaxonomyComparisonBase( gene_tree );
 205                 switch ( comp_base ) {
 206                     case SCIENTIFIC_NAME:
 207                         try {
 208                             PhylogenyMethods
 209                                     .transferNodeNameToField( species_tree,
 210                                                               PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
 211                                                               true );
 212                         }
 213                         catch ( final PhyloXmlDataFormatException e ) {
 214                             ForesterUtil.fatalError( gsdi.PRG_NAME,
 215                                                      "Failed to transfer general node name to scientific name, in ["
 216                                                              + species_tree_file + "]: " + e.getMessage() );
 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                             ForesterUtil.fatalError( gsdi.PRG_NAME,
 228                                                      "Failed to transfer general node name to taxonomy code, in ["
 229                                                              + species_tree_file + "]: " + e.getMessage() );
 230                         }
 231                         break;
 232                     case ID:
 233                         try {
 234                             PhylogenyMethods.transferNodeNameToField( species_tree,
 235                                                                       PhylogenyMethods.PhylogenyNodeField.TAXONOMY_ID,
 236                                                                       true );
 237                         }
 238                         catch ( final PhyloXmlDataFormatException e ) {
 239                             ForesterUtil.fatalError( gsdi.PRG_NAME,
 240                                                      "Failed to transfer general node name to taxonomy id, in ["
 241                                                              + species_tree_file + "]: " + e.getMessage() );
 242                         }
 243                         break;
 244                     default:
 245                         ForesterUtil.fatalError( gsdi.PRG_NAME, "unable to determine comparison base" );
 246                 }
 247             }
 248         }
 249         catch ( final IOException e ) {
 250             ForesterUtil.fatalError( gsdi.PRG_NAME, "Failed to read species tree from [" + species_tree_file + "]: "
 251                     + e.getMessage() );
 252         }
 253         gene_tree.setRooted( true );
 254         species_tree.setRooted( true );
 255         if ( !gene_tree.isCompletelyBinary() ) {
 256             log_writer.println( "User Error: gene tree is not completely binary" );
 257             log_writer.close();
 258             ForesterUtil.fatalError( gsdi.PRG_NAME, "gene tree is not completely binary" );
 259         }
 260         if ( base_algorithm != BASE_ALGORITHM.GSDI ) {
 261             if ( !species_tree.isCompletelyBinary() ) {
 262                 log_writer.println( "User Error: species tree is not completely binary, use GSDI instead" );
 263                 log_writer.close();
 264                 ForesterUtil.fatalError( gsdi.PRG_NAME, "species tree is not completely binary, use GSDI instead" );
 265             }
 266         }
 267         // For timing.
 268         // gene_tree = Helper.createBalancedTree( 10 );
 269         // species_tree = Helper.createBalancedTree( 13 );
 270         // species_tree = Helper.createUnbalancedTree( 1024 );
 271         // gene_tree = Helper.createUnbalancedTree( 8192 );
 272         // species_tree = gene_tree.copyTree();
 273         // gene_tree = species_tree.copyTree();
 274         // Helper.numberSpeciesInOrder( species_tree );
 275         // Helper.numberSpeciesInOrder( gene_tree );
 276         // Helper.randomizeSpecies( 1, 8192, gene_tree );
 277         // Helper.intervalNumberSpecies( gene_tree, 4096 );
 278         // Helper.numberSpeciesInDescOrder( gene_tree );
 279         log_writer.println( PRG_NAME + " - " + PRG_DESC );
 280         log_writer.println( "  version         : " + PRG_VERSION );
 281         log_writer.println( "  date            : " + PRG_DATE );
 282         log_writer.println( "  forester version: " + ForesterConstants.FORESTER_VERSION );
 283         log_writer.println( "Start time: " + new SimpleDateFormat( "yyyyMMdd HH:mm:ss" ).format( new Date() ) );
 284         log_writer.println( "Gene tree file: " + gene_tree_file.getCanonicalPath() );
 285         log_writer.println( "Gene tree name: "
 286                 + ( ForesterUtil.isEmpty( gene_tree.getName() ) ? "" : gene_tree.getName() ) );
 287         log_writer.println( "Species tree file: " + species_tree_file.getCanonicalPath() );
 288         log_writer.println( "Species tree name: "
 289                 + ( ForesterUtil.isEmpty( species_tree.getName() ) ? "" : gene_tree.getName() ) );
 290         System.out.println();
 291         SDI sdi = null;
 292         final long start_time = new Date().getTime();
 293         try {
 294             if ( base_algorithm == BASE_ALGORITHM.GSDI ) {
 295                 System.out.println();
 296                 System.out.println( "Use most parsimonous duplication model: " + most_parsimonous_duplication_model );
 297                 System.out.println( "Allow stripping of gene tree nodes    : " + allow_stripping_of_gene_tree );
 298                 log_writer.println( "Use most parsimonous duplication model: " + most_parsimonous_duplication_model );
 299                 log_writer.write( "Allow stripping of gene tree nodes    : " + allow_stripping_of_gene_tree );
 300                 log_writer.flush();
 301                 sdi = new GSDI( gene_tree,
 302                                 species_tree,
 303                                 most_parsimonous_duplication_model,
 304                                 allow_stripping_of_gene_tree,
 305                                 true );
 306             }
 307             else {
 308                 System.out.println();
 309                 System.out.println( "Using SDIse algorithm" );
 310                 log_writer.println( "Using SDIse algorithm" );
 311                 log_writer.flush();
 312                 sdi = new SDIse( gene_tree, species_tree );
 313             }
 314         }
 315         catch ( final SdiException e ) {
 316             log_writer.println( "User Error: " + e.getLocalizedMessage() );
 317             log_writer.close();
 318             ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
 319         }
 320         catch ( final IOException e ) {
 321             log_writer.println( "Error: " + e );
 322             log_writer.close();
 323             ForesterUtil.fatalError( PRG_NAME, e.toString() );
 324         }
 325         catch ( final Exception e ) {
 326             log_writer.println( "Error: " + e );
 327             log_writer.close();
 328             e.printStackTrace();
 329             System.exit( -1 );
 330         }
 331         System.out.println();
 332         System.out.println( "Running time (excluding I/O): " + ( new Date().getTime() - start_time ) + "ms" );
 333         log_writer.println( "Running time (excluding I/O): " + ( new Date().getTime() - start_time ) + "ms" );
 334         try {
 335             final PhylogenyWriter writer = new PhylogenyWriter();
 336             writer.toPhyloXML( out_file, gene_tree, 0 );
 337         }
 338         catch ( final IOException e ) {
 339             ForesterUtil.fatalError( PRG_NAME, "Failed to write to [" + out_file + "]: " + e.getMessage() );
 340         }
 341         System.out.println();
 342         System.out.println( "Wrote resulting gene tree to: " + out_file );
 343         System.out.println();
 344         log_writer.println( "Wrote resulting gene tree to: " + out_file );
 345         if ( base_algorithm == BASE_ALGORITHM.SDI ) {
 346             sdi.computeMappingCostL();
 347             System.out.println( "Mapping cost                    : " + sdi.computeMappingCostL() );
 348             log_writer.println( "Mapping cost                    : " + sdi.computeMappingCostL() );
 349         }
 350         System.out.println( "Number of duplications          : " + sdi.getDuplicationsSum() );
 351         log_writer.println( "Number of duplications          : " + sdi.getDuplicationsSum() );
 352         if ( ( base_algorithm == BASE_ALGORITHM.GSDI ) ) {
 353             final GSDI gsdi = ( GSDI ) sdi;
 354             final File species_tree_used_file = new File( out_file + SUFFIX_FOR_SPECIES_TREE_USED );
 355             try {
 356                 final PhylogenyWriter writer = new PhylogenyWriter();
 357                 writer.toPhyloXML( species_tree_used_file, gsdi.getSpeciesTree(), 0 );
 358             }
 359             catch ( final IOException e ) {
 360                 ForesterUtil.fatalError( PRG_NAME,
 361                                          "Failed to write to [" + species_tree_used_file + "]: " + e.getMessage() );
 362             }
 363             System.out.println();
 364             System.out.println( "Wrote used species tree to: " + species_tree_used_file );
 365             System.out.println();
 366             log_writer.println( "Wrote used species tree to: " + species_tree_used_file );
 367             if ( !most_parsimonous_duplication_model ) {
 368                 final int duplications = gsdi.getSpeciationOrDuplicationEventsSum();
 369                 System.out.println( "Number of potential duplications: " + duplications );
 370                 log_writer.println( "Number of potential duplications: " + duplications );
 371             }
 372             final int spec = gsdi.getSpeciationsSum();
 373             System.out.println( "Number of speciations            : " + spec );
 374             log_writer.println( "Number of speciations            : " + spec );
 375             for( final PhylogenyNode n : gsdi.getMappedExternalSpeciesTreeNodes() ) {
 376                 System.out.println( n.toString() );
 377             }
 378         }
 379         System.out.println();
 380         log_writer.close();
 381         //  some stat on gene tree:
 382         //      filename, name
 383         //      number of external nodes, strppided nodes
 384         //  some stats on sepcies tree, external nodes,
 385         //  filename, name
 386         //  internal nodes
 387         //  how many of which are polytomies
 388         //wrote log file to
 389         //  if ( allow_stripping_of_gene_tree ) {
 390         //      stripped x nodes, y external nodes remain
 391         //  }
 392     }
 393
 394     private static void print_help() {
 395         System.out.println( "Usage: " + gsdi.PRG_NAME
 396                 + " [-options] <gene tree in phyloXML format> <species tree in phyloXML format> [outfile]" );
 397         System.out.println();
 398         System.out.println( "Options:" );
 399         //    System.out.println( " -" + gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION + ": to allow stripping of gene tree nodes without a matching species in the species tree (writes list of stripped nodes to " + );
 400         System.out.println( " -" + gsdi.SDI_OPTION + ": to use SDI algorithm instead of GSDI algorithm" );//TODO gsdi.ALLOW_STRIPPING_OF_GENE_TREE_OPTION not allowed
 401         System.out.println( " -" + gsdi.MOST_PARSIMONIOUS_OPTION
 402                 + ": use most parimonious duplication model for GSDI: " );
 403         System.out.println( "     assign nodes as speciations which would otherwise be assiged" );
 404         System.out.println( "     as unknown because of polytomies in the species tree" );
 405         System.out.println( " -" + gsdi.GUESS_FORMAT_OF_SPECIES_TREE
 406                 + ": to allow species tree in other formats than phyloXML (i.e. Newick, NHX, Nexus)" );
 407         System.out.println();
 408         System.out.println( "Species tree:" );
 409         System.out
 410                 .println( " In phyloXML format (unless option -"
 411                         + gsdi.GUESS_FORMAT_OF_SPECIES_TREE
 412                         + " is used, in which case the species matching between gene tree and species tree must be via scientific name), with taxonomy data in appropriate fields" );
 413         System.out.println();
 414         System.out.println( "Gene tree:" );
 415         System.out.println( " In phyloXM format, with taxonomy and sequence data in appropriate fields" );
 416         System.out.println();
 417         System.out.println( "Example:" );
 418         //  System.out.println( "gsdi
 419         //    System.out.println();
 420         System.out.println( "Note -- GSDI algorithm is under development" );
 421         System.out.println();
 422     }
 423 }