forester/java/src/org/forester/application/sdi_r.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
  25
  26 package org.forester.application;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.util.ArrayList;
  31 import java.util.Date;
  32 import java.util.List;
  33
  34 import org.forester.io.parsers.PhylogenyParser;
  35 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
  36 import org.forester.io.writers.PhylogenyWriter;
  37 import org.forester.phylogeny.Phylogeny;
  38 import org.forester.phylogeny.PhylogenyMethods;
  39 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  40 import org.forester.phylogeny.factories.PhylogenyFactory;
  41 import org.forester.sdi.SDIR;
  42 import org.forester.util.CommandLineArguments;
  43 import org.forester.util.ForesterUtil;
  44
  45 public class sdi_r {
  46
  47     final static private String HELP_OPTION_1           = "help";
  48     final static private String HELP_OPTION_2           = "h";
  49     final static private String MIN_MAPPING_COST_OPTION = "ml";
  50     final static private String MIN_DUPS_OPTION         = "md";
  51     final static private String MIN_HEIGHT_OPTION       = "mh";
  52     final static private String PRG_NAME                = "sdi_r";
  53     final static private String PRG_VERSION             = "1.11";
  54     final static private String PRG_DATE                = "2009.06.19";
  55     final static private String E_MAIL                  = "czmasek@burnham.org";
  56     final static private String WWW                     = "www.phylosoft.org";
  57     // How many resulting trees "main" should return/display.
  58     private final static int    TREES_TO_RETURN         = 5;
  59
  60     public static void main( final String args[] ) {
  61         ForesterUtil.printProgramInformation( PRG_NAME, PRG_VERSION, PRG_DATE, E_MAIL, WWW );
  62         CommandLineArguments cla = null;
  63         try {
  64             cla = new CommandLineArguments( args );
  65         }
  66         catch ( final Exception e ) {
  67             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  68         }
  69         if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length == 0 ) ) {
  70             printHelp();
  71             System.exit( 0 );
  72         }
  73         if ( ( args.length < 3 ) || ( cla.getNumberOfNames() != 2 ) ) {
  74             System.out.println();
  75             System.out.println( "[" + PRG_NAME + "] incorrect number of arguments" );
  76             System.out.println();
  77             printHelp();
  78             System.exit( -1 );
  79         }
  80         final List<String> allowed_options = new ArrayList<String>();
  81         allowed_options.add( MIN_MAPPING_COST_OPTION );
  82         allowed_options.add( MIN_DUPS_OPTION );
  83         allowed_options.add( MIN_HEIGHT_OPTION );
  84         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  85         if ( dissallowed_options.length() > 0 ) {
  86             ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  87         }
  88         final File outfile = new File( "sdir_outfile.xml" );
  89         if ( outfile.exists() ) {
  90             ForesterUtil.fatalError( PRG_NAME, "outfile \"" + outfile + "\" already exists" );
  91         }
  92         final File gene_tree_file = cla.getFile( 0 );
  93         final File species_tree_file = cla.getFile( 1 );
  94         boolean minimize_cost = false;
  95         if ( cla.isOptionSet( MIN_MAPPING_COST_OPTION ) ) {
  96             minimize_cost = true;
  97         }
  98         boolean minimize_sum_of_dup = false;
  99         if ( cla.isOptionSet( MIN_DUPS_OPTION ) ) {
 100             minimize_sum_of_dup = true;
 101         }
 102         boolean minimize_height = false;
 103         if ( cla.isOptionSet( MIN_HEIGHT_OPTION ) ) {
 104             minimize_height = true;
 105         }
 106         int r = 0;
 107         Phylogeny[] gene_trees = null;
 108         Phylogeny species_tree = null;
 109         if ( minimize_cost && minimize_sum_of_dup ) {
 110             minimize_sum_of_dup = false;
 111         }
 112         final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 113         try {
 114             final PhylogenyParser pp = new PhyloXmlParser();
 115             species_tree = factory.create( species_tree_file, pp )[ 0 ];
 116         }
 117         catch ( final IOException e ) {
 118             ForesterUtil.fatalError( PRG_NAME, "failed to read species tree [" + species_tree_file + "]: "
 119                     + e.getLocalizedMessage() );
 120         }
 121         if ( !species_tree.isRooted() ) {
 122             ForesterUtil.fatalError( PRG_NAME, "species tree [" + species_tree_file + "] is not rooted" );
 123         }
 124         try {
 125             final PhylogenyParser pp = new PhyloXmlParser();
 126             gene_trees = factory.create( gene_tree_file, pp );
 127         }
 128         catch ( final IOException e ) {
 129             ForesterUtil.fatalError( PRG_NAME, "failed to read gene trees [" + gene_tree_file + "]: "
 130                     + e.getLocalizedMessage() );
 131         }
 132         // Removes from gene_tree all species not found in species_tree.
 133         int gene_tree_counter = 0;
 134         final List<Phylogeny> all_result_trees = new ArrayList<Phylogeny>();
 135         for( final Phylogeny gene_tree : gene_trees ) {
 136             r = PhylogenyMethods.taxonomyBasedDeletionOfExternalNodes( species_tree, gene_tree );
 137             ForesterUtil.programMessage( PRG_NAME, "Removed " + r + " external nodes from gene tree" );
 138             final SDIR sdiunrooted = new SDIR();
 139             final long start_time = new Date().getTime();
 140             Phylogeny[] result_trees = null;
 141             try {
 142                 result_trees = sdiunrooted.infer( gene_tree,
 143                                                   species_tree,
 144                                                   minimize_cost,
 145                                                   minimize_sum_of_dup,
 146                                                   minimize_height,
 147                                                   true,
 148                                                   sdi_r.TREES_TO_RETURN );
 149             }
 150             catch ( final Exception e ) {
 151                 ForesterUtil.fatalError( PRG_NAME, e.getLocalizedMessage() );
 152             }
 153             final long time_req = new Date().getTime() - start_time;
 154             if ( minimize_cost ) {
 155                 ForesterUtil.programMessage( PRG_NAME, "Rooted by minimizing mapping cost L" );
 156                 if ( minimize_height ) {
 157                     ForesterUtil.programMessage( PRG_NAME,
 158                                                  "Selected tree(s) with minimal height out of resulting trees" );
 159                 }
 160                 ForesterUtil.programMessage( PRG_NAME, "Number differently rooted trees minimizing criterion  : "
 161                         + sdiunrooted.getCount() );
 162                 ForesterUtil.programMessage( PRG_NAME, "Minimal cost                                          : "
 163                         + sdiunrooted.getMinimalMappingCost() );
 164                 ForesterUtil.programMessage( PRG_NAME, "Minimal duplications                                  : "
 165                         + sdiunrooted.getMinimalDuplications() );
 166                 if ( minimize_height ) {
 167                     ForesterUtil.programMessage( PRG_NAME, "Phylogeny height                                      : "
 168                             + ForesterUtil.FORMATTER_06.format( sdiunrooted.getMinimalTreeHeight() ) );
 169                     ForesterUtil.programMessage( PRG_NAME, "Difference in subtree heights                         : "
 170                             + ForesterUtil.FORMATTER_06.format( sdiunrooted.getMinimalDiffInSubTreeHeights() ) );
 171                 }
 172             }
 173             else if ( minimize_sum_of_dup ) {
 174                 ForesterUtil.programMessage( PRG_NAME, "Rooted by minimizing sum of duplications" );
 175                 if ( minimize_height ) {
 176                     ForesterUtil.programMessage( PRG_NAME,
 177                                                  "Selected tree(s) with minimal height out of resulting trees" );
 178                 }
 179                 ForesterUtil.programMessage( PRG_NAME, "Number differently rooted trees minimizing criterion        : "
 180                         + sdiunrooted.getCount() );
 181                 ForesterUtil.programMessage( PRG_NAME, "Minimal duplications                                        : "
 182                         + sdiunrooted.getMinimalDuplications() );
 183                 if ( minimize_height ) {
 184                     ForesterUtil.programMessage( PRG_NAME,
 185                                                  "Phylogeny height                                            : "
 186                                                          + ForesterUtil.FORMATTER_06.format( sdiunrooted
 187                                                                  .getMinimalTreeHeight() ) );
 188                     ForesterUtil.programMessage( PRG_NAME,
 189                                                  "Difference in subtree heights                               : "
 190                                                          + ForesterUtil.FORMATTER_06.format( sdiunrooted
 191                                                                  .getMinimalDiffInSubTreeHeights() ) );
 192                 }
 193             }
 194             else if ( minimize_height ) {
 195                 ForesterUtil.programMessage( PRG_NAME, "Rooted by minimizing tree height (midpoint rooting)." );
 196                 ForesterUtil.programMessage( PRG_NAME, "Minimal tree height                  : "
 197                         + ForesterUtil.FORMATTER_06.format( sdiunrooted.getMinimalTreeHeight() ) );
 198                 ForesterUtil.programMessage( PRG_NAME, "Minimal difference in subtree heights: "
 199                         + ForesterUtil.FORMATTER_06.format( sdiunrooted.getMinimalDiffInSubTreeHeights() ) );
 200                 ForesterUtil.programMessage( PRG_NAME, "Duplications in midpoint rooted tree : "
 201                         + sdiunrooted.getMinimalDuplications() );
 202             }
 203             else {
 204                 ForesterUtil.programMessage( PRG_NAME, "No (re) rooting was performed." );
 205                 ForesterUtil.programMessage( PRG_NAME, "Duplications in tree: " + sdiunrooted.getMinimalDuplications() );
 206             }
 207             ForesterUtil.programMessage( PRG_NAME, "Time requirement (minus I/O)                          : "
 208                     + time_req + "ms" );
 209             for( int i = 0; i < result_trees.length; ++i ) {
 210                 final String name = result_trees[ i ].getName();
 211                 if ( ForesterUtil.isEmpty( name ) ) {
 212                     result_trees[ i ].setName( "SDIR result [gene tree + " + gene_tree_counter + "]" + " " + i );
 213                 }
 214                 else {
 215                     result_trees[ i ].setName( name + " SDIR result [gene tree + " + gene_tree_counter + "]" + " " + i );
 216                 }
 217                 all_result_trees.add( result_trees[ i ] );
 218             }
 219             ++gene_tree_counter;
 220         } // for( final Phylogeny gene_tree : gene_trees )
 221         try {
 222             final PhylogenyWriter w = new PhylogenyWriter();
 223             w.toPhyloXML( outfile, all_result_trees, 0, ForesterUtil.LINE_SEPARATOR );
 224         }
 225         catch ( final IOException e ) {
 226             ForesterUtil.fatalError( PRG_NAME, "failure to write output to [" + outfile + "]: "
 227                     + e.getLocalizedMessage() );
 228         }
 229         ForesterUtil.programMessage( PRG_NAME, "Wrote: " + outfile );
 230         ForesterUtil.programMessage( PRG_NAME, "OK." );
 231     }
 232
 233     private static void printHelp() {
 234         System.out.println( "Usage: " + PRG_NAME
 235                 + " <options> <gene tree(s) in phyloXML format> <species tree in phyloXML format>\"" );
 236         System.out.println( "\nOptions:" );
 237         System.out.println( " -" + MIN_MAPPING_COST_OPTION
 238                 + " to root by minimizing the mapping cost L (and also the sum of duplications)" );
 239         System.out.println( " -" + MIN_DUPS_OPTION + " to root by minimizing the sum of duplications" );
 240         System.out.println( " -" + MIN_HEIGHT_OPTION
 241                 + " to root by minimizing tree height (can be used together with -" + MIN_MAPPING_COST_OPTION + " or -"
 242                 + MIN_DUPS_OPTION + ")" );
 243         System.out.println( "" );
 244     }
 245 }