forester/java/src/org/forester/application/surf_paup.java

   1 // $Id:
   2 //
   3 // FORESTER -- software libraries and applications
   4 // for evolutionary biology research and applications.
   5 //
   6 // Copyright (C) 2008-2009 Christian M. Zmasek
   7 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   8 // All rights reserved
   9 //
  10 // This library is free software; you can redistribute it and/or
  11 // modify it under the terms of the GNU Lesser General Public
  12 // License as published by the Free Software Foundation; either
  13 // version 2.1 of the License, or (at your option) any later version.
  14 //
  15 // This library is distributed in the hope that it will be useful,
  16 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18 // Lesser General Public License for more details.
  19 //
  20 // You should have received a copy of the GNU Lesser General Public
  21 // License along with this library; if not, write to the Free Software
  22 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  23 //
  24 // Contact: phylosoft @ gmail . com
  25 // WWW: www.phylosoft.org
  26
  27 package org.forester.application;
  28
  29 import java.io.File;
  30 import java.io.IOException;
  31 import java.util.ArrayList;
  32 import java.util.List;
  33
  34 import org.forester.evoinference.matrix.character.CharacterStateMatrix;
  35 import org.forester.evoinference.matrix.character.CharacterStateMatrix.BinaryStates;
  36 import org.forester.evoinference.matrix.character.CharacterStateMatrix.Format;
  37 import org.forester.io.parsers.nexus.NexusCharactersParser;
  38 import org.forester.io.parsers.nexus.NexusPhylogeniesParser;
  39 import org.forester.io.parsers.nexus.PaupLogParser;
  40 import org.forester.phylogeny.Phylogeny;
  41 import org.forester.phylogeny.PhylogenyNode;
  42 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  43 import org.forester.phylogeny.factories.PhylogenyFactory;
  44 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  45 import org.forester.surfacing.DomainParsimonyCalculator;
  46 import org.forester.surfacing.SurfacingUtil;
  47 import org.forester.util.CommandLineArguments;
  48 import org.forester.util.ForesterUtil;
  49
  50 public class surf_paup {
  51
  52     final static private String PRG_VERSION   = "0.90";
  53     final static private String PRG_DATE      = "2008.03.28";
  54     final static private String E_MAIL        = "czmasek@burnham.org";
  55     final static private String WWW           = "www.phylosoft.org/forester/applications/surfacing";
  56     final static private String HELP_OPTION_1 = "help";
  57     final static private String HELP_OPTION_2 = "h";
  58     private static final String PRG_NAME      = "surf_paup";
  59
  60     public static void main( final String args[] ) {
  61         ForesterUtil.printProgramInformation( PRG_NAME, PRG_VERSION, PRG_DATE, E_MAIL, WWW );
  62         final List<String> allowed_options = new ArrayList<String>();
  63         allowed_options.add( HELP_OPTION_1 );
  64         allowed_options.add( HELP_OPTION_2 );
  65         if ( ( args.length < 2 ) ) {
  66             printHelp();
  67             System.exit( -1 );
  68         }
  69         CommandLineArguments cla = null;
  70         try {
  71             cla = new CommandLineArguments( args );
  72         }
  73         catch ( final Exception e ) {
  74             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  75         }
  76         if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) ) {
  77             printHelp();
  78             System.exit( 0 );
  79         }
  80         if ( cla.getNumberOfNames() != 3 ) {
  81             printHelp();
  82             System.exit( -1 );
  83         }
  84         final File surfacing_nexus_outfile = cla.getFile( 0 );
  85         final File paup_log_file = cla.getFile( 1 );
  86         final String outfile_name = cla.getFile( 2 ).toString();
  87         final NexusCharactersParser nex_char_parser = new NexusCharactersParser();
  88         try {
  89             nex_char_parser.setSource( surfacing_nexus_outfile );
  90             nex_char_parser.parse();
  91         }
  92         catch ( final IOException e ) {
  93             ForesterUtil.fatalError( PRG_NAME, "problem with parsing character labels from  ["
  94                     + surfacing_nexus_outfile + "]: " + e.getMessage() );
  95             e.printStackTrace();
  96         }
  97         final String[] labels = nex_char_parser.getCharStateLabels();
  98         ForesterUtil.programMessage( PRG_NAME, "read in " + labels.length + " character labels" );
  99         final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 100         final NexusPhylogeniesParser phylogeny_parser = new NexusPhylogeniesParser();
 101         Phylogeny[] phylogenies = null;
 102         try {
 103             phylogenies = factory.create( surfacing_nexus_outfile, phylogeny_parser );
 104         }
 105         catch ( final IOException e ) {
 106             ForesterUtil.fatalError( PRG_NAME,
 107                                      "problem with parsing phylogeny [" + surfacing_nexus_outfile + "]: "
 108                                              + e.getMessage() );
 109             e.printStackTrace();
 110         }
 111         if ( phylogenies.length != 1 ) {
 112             ForesterUtil.fatalError( PRG_NAME, "failed to parse one phylogeny from [" + surfacing_nexus_outfile
 113                     + "], got " + phylogenies.length + " instead" );
 114         }
 115         final Phylogeny phylogeny = phylogenies[ 0 ];
 116         if ( !phylogeny.isRooted() ) {
 117             ForesterUtil.fatalError( PRG_NAME, "phylogeny from [" + surfacing_nexus_outfile + "] is not rooted" );
 118         }
 119         postOrderRelabelInternalNodes( phylogeny, phylogeny.getNumberOfExternalNodes() + 1 );
 120         CharacterStateMatrix<BinaryStates> matrix = null;
 121         final PaupLogParser paup_log_parser = new PaupLogParser();
 122         try {
 123             paup_log_parser.setSource( paup_log_file );
 124             matrix = paup_log_parser.parse();
 125         }
 126         catch ( final IOException e ) {
 127             ForesterUtil.fatalError( PRG_NAME,
 128                                      "failed to parse matrix from  [" + paup_log_file + "]: " + e.getMessage() );
 129         }
 130         ForesterUtil.programMessage( PRG_NAME,
 131                                      "read in character state matrix of size " + matrix.getNumberOfIdentifiers() + "x"
 132                                              + matrix.getNumberOfCharacters() );
 133         final DomainParsimonyCalculator domain_parsimony = DomainParsimonyCalculator.createInstance( phylogeny );
 134         domain_parsimony.executeOnGivenBinaryStatesMatrix( matrix, labels );
 135         final String sep = ForesterUtil.LINE_SEPARATOR + "###################" + ForesterUtil.LINE_SEPARATOR;
 136         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossMatrix(),
 137                                          outfile_name + "_paup_gl",
 138                                          Format.FORESTER );
 139         SurfacingUtil.writeMatrixToFile( domain_parsimony.getGainLossCountsMatrix(),
 140                                          outfile_name + "_paup_glc",
 141                                          Format.FORESTER );
 142         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 143                                                            CharacterStateMatrix.GainLossStates.GAIN,
 144                                                            outfile_name + "_paup_gains",
 145                                                            sep,
 146                                                            ForesterUtil.LINE_SEPARATOR,
 147                                                            null );
 148         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(),
 149                                                            CharacterStateMatrix.GainLossStates.LOSS,
 150                                                            outfile_name + "_paup_losses",
 151                                                            sep,
 152                                                            ForesterUtil.LINE_SEPARATOR,
 153                                                            null );
 154         SurfacingUtil.writeBinaryStatesMatrixAsListToFile( domain_parsimony.getGainLossMatrix(), null, outfile_name
 155                 + "_paup_present", sep, ForesterUtil.LINE_SEPARATOR, null );
 156         final String date_time = ForesterUtil.getCurrentDateTime();
 157         SurfacingUtil.preparePhylogeny( phylogeny, domain_parsimony, date_time, "parsimony (paup)", "paup_"
 158                 + outfile_name, "" );
 159         SurfacingUtil.writePhylogenyToFile( phylogeny, outfile_name + "_paup.xml" );
 160         ForesterUtil.programMessage( PRG_NAME, "OK" );
 161     }
 162
 163     final private static void postOrderRelabelInternalNodes( final Phylogeny phylogeny, final int starting_number ) {
 164         int i = starting_number;
 165         for( final PhylogenyNodeIterator it = phylogeny.iteratorPostorder(); it.hasNext(); ) {
 166             final PhylogenyNode node = it.next();
 167             if ( !node.isExternal() ) {
 168                 node.setName( String.valueOf( i++ ) );
 169             }
 170         }
 171     }
 172
 173     private static void printHelp() {
 174         System.out.println();
 175         System.out.println( "Usage:" );
 176         System.out.println();
 177         System.out
 178                 .println( "% java  -cp forester.jar org.forester.applications."
 179                         + PRG_NAME
 180                         + " <surfacing nexus outfile with character labels and tree> <paup log file with reconstructed states matrix> <outfile name base>" );
 181         System.out.println();
 182     }
 183 }