forester/java/src/org/forester/application/rio.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2017 Christian M. Zmasek
   6 // All rights reserved
   7 //
   8 // This library is free software; you can redistribute it and/or
   9 // modify it under the terms of the GNU Lesser General Public
  10 // License as published by the Free Software Foundation; either
  11 // version 2.1 of the License, or (at your option) any later version.
  12 //
  13 // This library is distributed in the hope that it will be useful,
  14 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  16 // Lesser General Public License for more details.
  17 //
  18 // You should have received a copy of the GNU Lesser General Public
  19 // License along with this library; if not, write to the Free Software
  20 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  21 //
  22 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
  23
  24 package org.forester.application;
  25
  26 import java.io.File;
  27 import java.io.FilenameFilter;
  28 import java.io.IOException;
  29 import java.util.ArrayList;
  30 import java.util.Arrays;
  31 import java.util.List;
  32
  33 import org.forester.rio.RIO;
  34 import org.forester.rio.RIO.REROOTING;
  35 import org.forester.rio.RIOUtil;
  36 import org.forester.sdi.SDIutil.ALGORITHM;
  37 import org.forester.util.CommandLineArguments;
  38 import org.forester.util.EasyWriter;
  39 import org.forester.util.ForesterUtil;
  40
  41 public class rio {
  42     //
  43
  44     public final static String  PRG_NAME                       = "rio";
  45     public final static String  PRG_VERSION                    = "5.900";
  46     public final static String  PRG_DATE                       = "170420";
  47     final static private String E_MAIL                         = "phyloxml@gmail.com";
  48     final static private String WWW                            = "https://sites.google.com/site/cmzmasek/home/software/forester";
  49     final static private String HELP_OPTION_1                  = "help";
  50     final static private String LOGFILE_SUFFIX                 = "_RIO_log.tsv";
  51     final static private String STRIPPED_SPECIES_TREE_SUFFIX   = "_RIO_sst.xml";
  52     final static private String ORTHO_OUTTABLE_SUFFIX          = "_RIO_orthologies.tsv";
  53     final static private String ORTHO_OUTTABLE_WITH_MAP_SUFFIX = "_RIO_orthologies_ext_map.tsv";
  54     final static private String OUT_MIN_DUP_GENE_TREE_SUFFIX   = "_RIO_gene_tree_min_dup_";
  55     final static private String OUT_MED_DUP_GENE_TREE_SUFFIX   = "_RIO_gene_tree_med_dup_";
  56     final static private String ORTHOLOG_GROUPS_SUFFIX         = "_RIO_ortholog_groups.tsv";
  57     final static private String HELP_OPTION_2                  = "h";
  58     final static private String GT_FIRST                       = "f";
  59     final static private String GT_LAST                        = "l";
  60     final static private String REROOTING_OPT                  = "r";
  61     final static private String OUTGROUP                       = "o";
  62     final static private String USE_SDIR                       = "s";
  63     final static private String GENE_TREES_SUFFIX_OPTION       = "g";
  64     final static private String MAPPINGS_DIR_OPTION            = "m";
  65     final static private String MAPPINGS_SUFFIX_OPTION         = "ms";
  66     final static private String MAPPINGS_SUFFIX_DEFAULT        = ".nim";
  67     final static private String ORTHOLOG_GROUPS_CUTOFF_OPTION  = "c";
  68     final static private String GENE_TREES_SUFFIX_DEFAULT      = ".mlt";
  69     final static private double ORTHOLOG_GROUPS_CUTOFF_DEFAULT = 0.5;
  70
  71     public static void main( final String[] args ) {
  72         ForesterUtil.printProgramInformation( PRG_NAME,
  73                                               "resampled inference of orthologs",
  74                                               PRG_VERSION,
  75                                               PRG_DATE,
  76                                               E_MAIL,
  77                                               WWW,
  78                                               ForesterUtil.getForesterLibraryInformation() );
  79         CommandLineArguments cla = null;
  80         try {
  81             cla = new CommandLineArguments( args );
  82         }
  83         catch ( final Exception e ) {
  84             ForesterUtil.fatalError( e.getMessage() );
  85         }
  86         if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length == 0 ) ) {
  87             printHelp();
  88         }
  89         if ( ( args.length < 3 ) || ( args.length > 11 ) || ( cla.getNumberOfNames() < 3 ) ) {
  90             System.out.println();
  91             System.out.println( "error: incorrect number of arguments" );
  92             System.out.println();
  93             printHelp();
  94         }
  95         final List<String> allowed_options = new ArrayList<String>();
  96         allowed_options.add( GT_FIRST );
  97         allowed_options.add( GT_LAST );
  98         allowed_options.add( REROOTING_OPT );
  99         allowed_options.add( OUTGROUP );
 100         allowed_options.add( USE_SDIR );
 101         allowed_options.add( GENE_TREES_SUFFIX_OPTION );
 102         allowed_options.add( ORTHOLOG_GROUPS_CUTOFF_OPTION );
 103         allowed_options.add( MAPPINGS_DIR_OPTION );
 104         allowed_options.add( MAPPINGS_SUFFIX_OPTION );
 105         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 106         if ( dissallowed_options.length() > 0 ) {
 107             ForesterUtil.fatalError( "unknown option(s): " + dissallowed_options );
 108         }
 109         final File gene_trees_file = cla.getFile( 0 );
 110         final boolean use_dir;
 111         File indir = null;
 112         File outdir = null;
 113         if ( gene_trees_file.isDirectory() ) {
 114             if ( !gene_trees_file.exists() ) {
 115                 ForesterUtil.fatalError( "gene trees directory \"" + gene_trees_file + "\" does not exist" );
 116             }
 117             if ( gene_trees_file.listFiles().length < 1 ) {
 118                 ForesterUtil.fatalError( "gene trees directory \"" + gene_trees_file + "\" is empty" );
 119             }
 120             use_dir = true;
 121             indir = gene_trees_file;
 122         }
 123         else {
 124             use_dir = false;
 125         }
 126         final File species_tree_file = cla.getFile( 1 );
 127         File orthology_outtable = null;
 128         if ( use_dir ) {
 129             outdir = cla.getFile( 2 );
 130         }
 131         else {
 132             orthology_outtable = cla.getFile( 2 );
 133         }
 134         File logfile;
 135         if ( use_dir ) {
 136             if ( ( cla.getNumberOfNames() < 4 ) ) {
 137                 System.out.println();
 138                 System.out.println( "error: incorrect number of arguments" );
 139                 System.out.println();
 140                 printHelp();
 141             }
 142             logfile = cla.getFile( 3 );
 143             if ( logfile.exists() ) {
 144                 ForesterUtil.fatalError( "\"" + logfile + "\" already exists" );
 145             }
 146         }
 147         else {
 148             if ( cla.getNumberOfNames() > 3 ) {
 149                 logfile = cla.getFile( 3 );
 150                 if ( logfile.exists() ) {
 151                     ForesterUtil.fatalError( "\"" + logfile + "\" already exists" );
 152                 }
 153             }
 154             else {
 155                 logfile = null;
 156             }
 157         }
 158         boolean sdir = false;
 159         if ( cla.isOptionSet( USE_SDIR ) ) {
 160             if ( cla.isOptionHasAValue( USE_SDIR ) ) {
 161                 ForesterUtil.fatalError( "no value allowed for -" + USE_SDIR );
 162             }
 163             sdir = true;
 164             if ( !use_dir && logfile != null ) {
 165                 ForesterUtil.fatalError( "no logfile output for SDIR algorithm" );
 166             }
 167         }
 168         String outgroup = null;
 169         if ( cla.isOptionSet( OUTGROUP ) ) {
 170             if ( sdir ) {
 171                 ForesterUtil.fatalError( "no outgroup option for SDIR algorithm" );
 172             }
 173             if ( use_dir ) {
 174                 ForesterUtil.fatalError( "no outgroup option for operating on gene trees directory" );
 175             }
 176             if ( !cla.isOptionHasAValue( OUTGROUP ) ) {
 177                 ForesterUtil.fatalError( "no value for -" + OUTGROUP );
 178             }
 179             outgroup = cla.getOptionValueAsCleanString( OUTGROUP );
 180         }
 181         REROOTING rerooting = REROOTING.BY_ALGORITHM;
 182         if ( cla.isOptionSet( REROOTING_OPT ) ) {
 183             if ( !cla.isOptionHasAValue( REROOTING_OPT ) ) {
 184                 ForesterUtil.fatalError( "no value for -" + REROOTING_OPT );
 185             }
 186             if ( sdir ) {
 187                 ForesterUtil.fatalError( "no re-rooting option for SDIR algorithm" );
 188             }
 189             final String rerooting_str = cla.getOptionValueAsCleanString( REROOTING_OPT ).toLowerCase();
 190             if ( rerooting_str.equals( "none" ) ) {
 191                 rerooting = REROOTING.NONE;
 192             }
 193             else if ( rerooting_str.equals( "midpoint" ) ) {
 194                 rerooting = REROOTING.MIDPOINT;
 195             }
 196             else if ( rerooting_str.equals( "outgroup" ) ) {
 197                 if ( use_dir ) {
 198                     ForesterUtil.fatalError( "no outgroup option for operating on gene trees directory" );
 199                 }
 200                 rerooting = REROOTING.OUTGROUP;
 201             }
 202             else {
 203                 ForesterUtil
 204                         .fatalError( "values for re-rooting are: 'none', 'midpoint', or 'outgroup' (minizming duplications is default)" );
 205             }
 206         }
 207         if ( ForesterUtil.isEmpty( outgroup ) && ( rerooting == REROOTING.OUTGROUP ) ) {
 208             ForesterUtil.fatalError( "selected re-rooting by outgroup, but outgroup not set" );
 209         }
 210         if ( !ForesterUtil.isEmpty( outgroup ) && ( rerooting != REROOTING.OUTGROUP ) ) {
 211             ForesterUtil.fatalError( "outgroup set, but selected re-rooting by other approach" );
 212         }
 213         int gt_first = RIO.DEFAULT_RANGE;
 214         int gt_last = RIO.DEFAULT_RANGE;
 215         if ( cla.isOptionSet( GT_FIRST ) ) {
 216             if ( sdir ) {
 217                 ForesterUtil.fatalError( "no gene tree range option for SDIR algorithm" );
 218             }
 219             if ( !cla.isOptionHasAValue( GT_FIRST ) ) {
 220                 ForesterUtil.fatalError( "no value for -" + GT_FIRST );
 221             }
 222             try {
 223                 gt_first = cla.getOptionValueAsInt( GT_FIRST );
 224             }
 225             catch ( final IOException e ) {
 226                 ForesterUtil.fatalError( "could not parse integer for -" + GT_FIRST + " option" );
 227             }
 228             if ( gt_first < 0 ) {
 229                 ForesterUtil.fatalError( "attempt to set index of first tree to analyze to: " + gt_first );
 230             }
 231         }
 232         if ( cla.isOptionSet( GT_LAST ) ) {
 233             if ( sdir ) {
 234                 ForesterUtil.fatalError( "no gene tree range option for SDIR algorithm" );
 235             }
 236             if ( !cla.isOptionHasAValue( GT_LAST ) ) {
 237                 ForesterUtil.fatalError( "no value for -" + GT_LAST );
 238             }
 239             try {
 240                 gt_last = cla.getOptionValueAsInt( GT_LAST );
 241             }
 242             catch ( final IOException e ) {
 243                 ForesterUtil.fatalError( "could not parse integer for -" + GT_LAST + " option" );
 244             }
 245             if ( gt_last < 0 ) {
 246                 ForesterUtil.fatalError( "attempt to set index of last tree to analyze to: " + gt_last );
 247             }
 248         }
 249         if ( ( ( gt_last != RIO.DEFAULT_RANGE ) && ( gt_first != RIO.DEFAULT_RANGE ) ) && ( ( gt_last < gt_first ) ) ) {
 250             ForesterUtil.fatalError( "attempt to set range (0-based) of gene to analyze to: from " + gt_first + " to "
 251                     + gt_last );
 252         }
 253         double ortholog_group_cutoff = ORTHOLOG_GROUPS_CUTOFF_DEFAULT;
 254         if ( cla.isOptionSet( ORTHOLOG_GROUPS_CUTOFF_OPTION ) ) {
 255             if ( sdir ) {
 256                 ForesterUtil.fatalError( "ortholog groups cutoff for SDIR algorithm" );
 257             }
 258             if ( !cla.isOptionHasAValue( ORTHOLOG_GROUPS_CUTOFF_OPTION ) ) {
 259                 ForesterUtil.fatalError( "no value for -" + ORTHOLOG_GROUPS_CUTOFF_OPTION );
 260             }
 261             try {
 262                 ortholog_group_cutoff = cla.getOptionValueAsDouble( ORTHOLOG_GROUPS_CUTOFF_OPTION );
 263             }
 264             catch ( final IOException e ) {
 265                 ForesterUtil.fatalError( "could not parse double for -" + ORTHOLOG_GROUPS_CUTOFF_OPTION + " option" );
 266             }
 267             if ( ortholog_group_cutoff < 0 ) {
 268                 ForesterUtil.fatalError( "attempt to set ortholog groups cutoff to: " + ortholog_group_cutoff );
 269             }
 270             if ( ortholog_group_cutoff > 1 ) {
 271                 ForesterUtil.fatalError( "attempt to set ortholog groups cutoff to: " + ortholog_group_cutoff );
 272             }
 273         }
 274         if ( !use_dir ) {
 275             ForesterUtil.fatalErrorIfFileNotReadable( gene_trees_file );
 276         }
 277         final String gene_trees_suffix;
 278         if ( cla.isOptionSet( GENE_TREES_SUFFIX_OPTION ) ) {
 279             if ( !use_dir ) {
 280                 ForesterUtil.fatalError( "no gene tree suffix option when operating on indivual gene trees" );
 281             }
 282             if ( !cla.isOptionHasAValue( GENE_TREES_SUFFIX_OPTION ) ) {
 283                 ForesterUtil.fatalError( "no value for -" + GENE_TREES_SUFFIX_OPTION );
 284             }
 285             gene_trees_suffix = cla.getOptionValueAsCleanString( GENE_TREES_SUFFIX_OPTION );
 286         }
 287         else {
 288             gene_trees_suffix = GENE_TREES_SUFFIX_DEFAULT;
 289         }
 290         final boolean perform_id_mapping;
 291         final File id_mapping_dir;
 292         if ( cla.isOptionSet( MAPPINGS_DIR_OPTION ) ) {
 293             id_mapping_dir = new File( cla.getOptionValue( MAPPINGS_DIR_OPTION ) );
 294             perform_id_mapping = true;
 295             if ( !use_dir ) {
 296                 ForesterUtil.fatalError( "no id mapping when operating on indivual gene trees" );
 297             }
 298             if ( !id_mapping_dir.exists() ) {
 299                 ForesterUtil.fatalError( "id mappings directory \"" + id_mapping_dir + "\" does not exist" );
 300             }
 301             if ( !id_mapping_dir.isDirectory() ) {
 302                 ForesterUtil.fatalError( "id mappings directory \"" + id_mapping_dir + "\" is not a directory" );
 303             }
 304             if ( id_mapping_dir.listFiles().length < 1 ) {
 305                 ForesterUtil.fatalError( "id mappings directory \"" + id_mapping_dir + "\" is empty" );
 306             }
 307         }
 308         else {
 309             id_mapping_dir = null;
 310             perform_id_mapping = false;
 311         }
 312         final String id_mapping_suffix;
 313         if ( cla.isOptionSet( MAPPINGS_SUFFIX_OPTION ) ) {
 314             if ( !use_dir ) {
 315                 ForesterUtil.fatalError( "no id mapping file suffix option when operating on indivual gene trees" );
 316             }
 317             if ( !perform_id_mapping ) {
 318                 ForesterUtil.fatalError( "no id mapping directory given" );
 319             }
 320             if ( !cla.isOptionHasAValue( MAPPINGS_SUFFIX_OPTION ) ) {
 321                 ForesterUtil.fatalError( "no value for -" + MAPPINGS_SUFFIX_OPTION );
 322             }
 323             id_mapping_suffix = cla.getOptionValueAsCleanString( MAPPINGS_SUFFIX_OPTION );
 324         }
 325         else {
 326             id_mapping_suffix = MAPPINGS_SUFFIX_DEFAULT;
 327         }
 328         ForesterUtil.fatalErrorIfFileNotReadable( species_tree_file );
 329         if ( !use_dir && orthology_outtable.exists() ) {
 330             ForesterUtil.fatalError( "\"" + orthology_outtable + "\" already exists" );
 331         }
 332         long time = 0;
 333         try {
 334             if ( use_dir ) {
 335                 System.out.println( "Gene trees in-dir                   :\t" + indir.getCanonicalPath() );
 336                 System.out.println( "Gene trees suffix                   :\t" + gene_trees_suffix );
 337             }
 338             else {
 339                 System.out.println( "Gene trees                          :\t" + gene_trees_file.getCanonicalPath() );
 340             }
 341             System.out.println( "Species tree                        :\t" + species_tree_file.getCanonicalPath() );
 342         }
 343         catch ( final IOException e ) {
 344             ForesterUtil.fatalError( e.getLocalizedMessage() );
 345         }
 346         if ( perform_id_mapping ) {
 347             try {
 348                 System.out.println( "Id mappings in-dir                  :\t" + id_mapping_dir.getCanonicalPath() );
 349             }
 350             catch ( IOException e ) {
 351                 ForesterUtil.fatalError( e.getLocalizedMessage() );
 352             }
 353             System.out.println( "Id mappings suffix                  :\t" + id_mapping_suffix );
 354         }
 355         if ( use_dir ) {
 356             System.out.println( "Out-dir                             :\t" + outdir );
 357         }
 358         else {
 359             System.out.println( "All vs all orthology results table  :\t" + orthology_outtable );
 360         }
 361         if ( logfile != null ) {
 362             System.out.println( "Logfile                             :\t" + logfile );
 363         }
 364         System.out.println( "Ortholog groups cutoff              :\t" + ortholog_group_cutoff );
 365         if ( gt_first != RIO.DEFAULT_RANGE ) {
 366             System.out.println( "First gene tree to analyze          :\t" + gt_first );
 367         }
 368         if ( gt_last != RIO.DEFAULT_RANGE ) {
 369             System.out.println( "Last gene tree to analyze           :\t" + gt_last );
 370         }
 371         String rerooting_str = "";
 372         switch ( rerooting ) {
 373             case BY_ALGORITHM: {
 374                 rerooting_str = "by minimizing duplications";
 375                 break;
 376             }
 377             case MIDPOINT: {
 378                 rerooting_str = "by midpoint method";
 379                 break;
 380             }
 381             case OUTGROUP: {
 382                 rerooting_str = "by outgroup: " + outgroup;
 383                 break;
 384             }
 385             case NONE: {
 386                 rerooting_str = "none";
 387                 break;
 388             }
 389         }
 390         System.out.println( "Re-rooting                          : \t" + rerooting_str );
 391         if ( !sdir ) {
 392             System.out.println( "Non binary species tree             :\tallowed" );
 393         }
 394         else {
 395             System.out.println( "Non binary species tree             :\tdisallowed" );
 396         }
 397         time = System.currentTimeMillis();
 398         final ALGORITHM algorithm;
 399         if ( sdir ) {
 400             algorithm = ALGORITHM.SDIR;
 401         }
 402         else {
 403             algorithm = ALGORITHM.GSDIR;
 404         }
 405         EasyWriter log = null;
 406         if ( use_dir ) {
 407             if ( outdir.exists() ) {
 408                 if ( !outdir.isDirectory() ) {
 409                     ForesterUtil.fatalError( PRG_NAME,
 410                                              "out-directory [" + outdir + "] already exists but is not a directory" );
 411                 }
 412             }
 413             else {
 414                 final boolean success = outdir.mkdirs();
 415                 if ( !success ) {
 416                     ForesterUtil.fatalError( PRG_NAME, "could not create out-directory [" + outdir + "]" );
 417                 }
 418             }
 419             final String species_tree_file_name = species_tree_file.getName();
 420             final File gene_trees_files[] = indir.listFiles( new FilenameFilter() {
 421
 422                 @Override
 423                 public boolean accept( final File dir, final String name ) {
 424                     return ( ( name.endsWith( gene_trees_suffix ) ) && !( name.equals( species_tree_file_name ) ) );
 425                 }
 426             } );
 427             if ( gene_trees_files.length < 1 ) {
 428                 ForesterUtil.fatalError( PRG_NAME,
 429                                          "in-directory [" + indir
 430                                                  + "] does not contain any gene tree files with suffix "
 431                                                  + gene_trees_suffix );
 432             }
 433             try {
 434                 log = ForesterUtil.createEasyWriter( logfile );
 435             }
 436             catch ( final IOException e ) {
 437                 ForesterUtil.fatalError( PRG_NAME, "could not create [" + logfile + "]" );
 438             }
 439             Arrays.sort( gene_trees_files );
 440             try {
 441                 log.print( "# program" );
 442                 log.print( "\t" );
 443                 log.print( PRG_NAME );
 444                 log.println();
 445                 log.print( "# version" );
 446                 log.print( "\t" );
 447                 log.print( PRG_VERSION );
 448                 log.println();
 449                 log.print( "# date" );
 450                 log.print( "\t" );
 451                 log.print( PRG_DATE );
 452                 log.println();
 453                 log.print( "# Algorithm " );
 454                 log.print( "\t" );
 455                 log.print( algorithm.toString() );
 456                 log.println();
 457                 log.print( "# Gene trees in-dir" );
 458                 log.print( "\t" );
 459                 log.print( indir.getCanonicalPath() );
 460                 log.println();
 461                 log.print( "# Gene trees suffix" );
 462                 log.print( "\t" );
 463                 log.print( gene_trees_suffix );
 464                 log.println();
 465                 log.print( "# Species tree" );
 466                 log.print( "\t" );
 467                 log.print( species_tree_file.getCanonicalPath() );
 468                 log.println();
 469                 log.print( "# Out-dir" );
 470                 log.print( "\t" );
 471                 log.print( outdir.getCanonicalPath() );
 472                 log.println();
 473                 log.print( "# Logfile" );
 474                 log.print( "\t" );
 475                 log.print( logfile.getCanonicalPath() );
 476                 log.println();
 477                 log.print( "# Ortholog groups cutoff" );
 478                 log.print( "\t" );
 479                 log.print( Double.toString( ortholog_group_cutoff ) );
 480                 log.println();
 481                 if ( gt_first != RIO.DEFAULT_RANGE ) {
 482                     log.print( "# First gene tree to analyze" );
 483                     log.print( "\t" );
 484                     log.print( Integer.toString( gt_first ) );
 485                     log.println();
 486                 }
 487                 if ( gt_last != RIO.DEFAULT_RANGE ) {
 488                     log.print( "# Last gene tree to analyze" );
 489                     log.print( "\t" );
 490                     log.print( Integer.toString( gt_last ) );
 491                     log.println();
 492                 }
 493                 log.print( "# Re-rooting" );
 494                 log.print( "\t" );
 495                 log.print( rerooting_str );
 496                 log.println();
 497                 log.print( "# Non binary species tree" );
 498                 log.print( "\t" );
 499                 if ( !sdir ) {
 500                     log.print( "allowed" );
 501                 }
 502                 else {
 503                     log.print( "disallowed" );
 504                 }
 505                 log.println();
 506                 log.println();
 507                 log.print( "NAME" );
 508                 log.print( "\t" );
 509                 log.print( "EXT NODES" );
 510                 log.print( "\t" );
 511                 log.print( ortholog_group_cutoff + " O GROUPS" );
 512                 log.print( "\t" );
 513                 log.print( "0.05 O GROUPS" );
 514                 log.print( "\t" );
 515                 log.print( "0.25 O GROUPS" );
 516                 log.print( "\t" );
 517                 log.print( "0.5 O GROUPS" );
 518                 log.print( "\t" );
 519                 log.print( "0.75 O GROUPS" );
 520                 log.print( "\t" );
 521                 log.print( "0.95 O GROUPS" );
 522                 log.print( "\t" );
 523                 if ( true ) { //TODO
 524                     log.print( "BEST TREE DUP" );
 525                     log.print( "\t" );
 526                 }
 527                 log.print( "MEDIAN DUP" );
 528                 log.print( "\t" );
 529                 log.print( "MEAN DUP" );
 530                 log.print( "\t" );
 531                 log.print( "MEAN DUP SD" );
 532                 log.print( "\t" );
 533                 log.print( "MIN DUP" );
 534                 log.print( "\t" );
 535                 log.print( "MAX DUP" );
 536                 log.print( "\t" );
 537                 log.print( "REMOVED EXT NODES" );
 538                 log.print( "\t" );
 539                 log.print( "N" );
 540                 log.println();
 541             }
 542             catch ( IOException e ) {
 543                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
 544             }
 545             int counter = 1;
 546             for( final File gf : gene_trees_files ) {
 547                 String outname = gf.getName();
 548                 System.out
 549                         .print( "\r                                                                                            " );
 550                 System.out.print( "\r" + counter + "/" + gene_trees_files.length + ": " + outname );
 551                 counter++;
 552                 if ( outname.indexOf( "." ) > 0 ) {
 553                     outname = outname.substring( 0, outname.lastIndexOf( "." ) );
 554                 }
 555                 try {
 556                     RIOUtil.executeAnalysis( gf,
 557                                              species_tree_file,
 558                                              new File( outdir.getCanonicalFile() + "/" + outname
 559                                                      + ORTHO_OUTTABLE_SUFFIX ),
 560                                              new File( outdir.getCanonicalFile() + "/" + outname
 561                                                      + ORTHO_OUTTABLE_WITH_MAP_SUFFIX ),
 562                                              new File( outdir.getCanonicalFile() + "/" + outname
 563                                                      + ORTHOLOG_GROUPS_SUFFIX ),
 564                                              new File( outdir.getCanonicalFile() + "/" + outname + LOGFILE_SUFFIX ),
 565                                              outgroup,
 566                                              rerooting,
 567                                              gt_first,
 568                                              gt_last,
 569                                              new File( outdir.getCanonicalFile() + "/" + outname
 570                                                      + STRIPPED_SPECIES_TREE_SUFFIX ),
 571                                              new File( outdir.getCanonicalFile() + "/" + outname
 572                                                      + OUT_MIN_DUP_GENE_TREE_SUFFIX ),
 573                                              new File( outdir.getCanonicalFile() + "/" + outname
 574                                                      + OUT_MED_DUP_GENE_TREE_SUFFIX ),
 575                                              true,
 576                                              algorithm,
 577                                              true,
 578                                              log,
 579                                              ortholog_group_cutoff,
 580                                              perform_id_mapping,
 581                                              id_mapping_dir,
 582                                              id_mapping_suffix );
 583                 }
 584                 catch ( IOException e ) {
 585                     ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
 586                 }
 587             }
 588             System.out
 589                     .print( "\r                                                                                        " );
 590             System.out.println();
 591         }
 592         else {
 593             String outname = orthology_outtable.toString();
 594             if ( outname.indexOf( "." ) > 0 ) {
 595                 outname = outname.substring( 0, outname.lastIndexOf( "." ) );
 596             }
 597             RIOUtil.executeAnalysis( gene_trees_file,
 598                                      species_tree_file,
 599                                      orthology_outtable,
 600                                      null,
 601                                      new File( outname + ORTHOLOG_GROUPS_SUFFIX ),
 602                                      logfile,
 603                                      outgroup,
 604                                      rerooting,
 605                                      gt_first,
 606                                      gt_last,
 607                                      new File( outname + STRIPPED_SPECIES_TREE_SUFFIX ),
 608                                      new File( outname + OUT_MIN_DUP_GENE_TREE_SUFFIX ),
 609                                      new File( outname + OUT_MED_DUP_GENE_TREE_SUFFIX ),
 610                                      algorithm == ALGORITHM.GSDIR,
 611                                      algorithm,
 612                                      false,
 613                                      null,
 614                                      ortholog_group_cutoff,
 615                                      false,
 616                                      null,
 617                                      null );
 618         }
 619         if ( !use_dir ) {
 620             time = System.currentTimeMillis() - time;
 621             System.out.println( "Time                                :\t" + time + "ms" );
 622         }
 623         else {
 624             try {
 625                 log.close();
 626             }
 627             catch ( IOException e ) {
 628                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
 629             }
 630             time = System.currentTimeMillis() - time;
 631             System.out.println( "Time                                :\t" + time + "ms" );
 632         }
 633         System.exit( 0 );
 634     }
 635
 636     private final static void printHelp() {
 637         System.out.println( "Usage" );
 638         System.out.println();
 639         System.out.println( PRG_NAME
 640                 + " [options] <gene trees infile> <species tree infile> <all vs all orthology table outfile> [logfile]" );
 641         System.out.println();
 642         System.out.println( PRG_NAME + " [options] <gene trees indir> <species tree infile> <outdir> <logfile>" );
 643         System.out.println();
 644         System.out.println();
 645         System.out.println( " Options" );
 646         System.out.println( "  -" + GT_FIRST + "=<first>     : first gene tree to analyze (0-based index)" );
 647         System.out.println( "  -" + GT_LAST + "=<last>      : last gene tree to analyze (0-based index)" );
 648         System.out.println( "  -" + ORTHOLOG_GROUPS_CUTOFF_OPTION
 649                 + "=<cutoff>    : cutoff value for ortholog groups (default: " + ORTHOLOG_GROUPS_CUTOFF_DEFAULT + ")" );
 650         System.out.println( "  -" + REROOTING_OPT
 651                 + "=<re-rooting>: re-rooting method for gene trees, possible values or 'none', 'midpoint'," );
 652         System.out.println( "                   or 'outgroup' (default: by minizming duplications)" );
 653         System.out.println( "  -" + OUTGROUP
 654                 + "=<outgroup>  : for rooting by outgroup, name of outgroup (external gene tree node)" );
 655         System.out.println( "  -" + USE_SDIR
 656                 + "             : to use SDIR instead of GSDIR (faster, but non-binary species trees are" );
 657         System.out.println( "                   disallowed, as are most options)" );
 658         System.out.println( "  -" + GENE_TREES_SUFFIX_OPTION
 659                 + "=<suffix>    : suffix for gene trees when operating on gene tree directories (default: "
 660                 + GENE_TREES_SUFFIX_DEFAULT + ")" );
 661         System.out.println( "  -" + MAPPINGS_DIR_OPTION + "=<dir>       : directory for id mapping files" );
 662         System.out.println( "  -" + MAPPINGS_SUFFIX_OPTION + "=<suffix>   : suffix for id mapping files (default: "
 663                 + MAPPINGS_SUFFIX_DEFAULT + ")" );
 664         System.out.println();
 665         System.out.println( " Formats" );
 666         System.out
 667                 .println( "  The gene trees, as well as the species tree, ideally are in phyloXML (www.phyloxml.org) format," );
 668         System.out
 669                 .println( "  but can also be in New Hamphshire (Newick) or Nexus format as long as species information can be" );
 670         System.out
 671                 .println( "  extracted from the gene names (e.g. \"HUMAN\" from \"BCL2_HUMAN\") and matched to a single species" );
 672         System.out.println( "  in the species tree." );
 673         System.out.println();
 674         System.out.println( " Examples" );
 675         System.out.println( "  rio -s gene_trees.nh species.xml outtable.tsv" );
 676         System.out.println( "  rio gene_trees.nh species.xml outtable.tsv log.txt" );
 677         System.out.println( "  rio -c=0.9 -f=10 -l=100 -r=none gene_trees.xml species.xml outtable.tsv log.txt" );
 678         System.out.println( "  rio -g=.xml gene_trees_dir species.xml out_dir log.tsv" );
 679         System.out.println( "  rio -g=.mlt -m=id_maps_dir -ms=.nim -c=0.8 gene_trees_dir species.xml out_dir log.tsv" );
 680         System.out.println( "  rio -m=id_maps_dir -c=0.8 gene_trees_dir species.xml out_dir log.tsv" );
 681         System.out.println();
 682         System.exit( -1 );
 683     }
 684 }