forester/java/src/org/forester/application/count_support.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.util.ArrayList;
  31 import java.util.Arrays;
  32 import java.util.List;
  33
  34 import org.forester.io.parsers.PhylogenyParser;
  35 import org.forester.io.parsers.util.ParserUtils;
  36 import org.forester.io.writers.PhylogenyWriter;
  37 import org.forester.phylogeny.Phylogeny;
  38 import org.forester.phylogeny.PhylogenyMethods;
  39 import org.forester.phylogeny.PhylogenyNode;
  40 import org.forester.phylogeny.data.Confidence;
  41 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  42 import org.forester.phylogeny.factories.PhylogenyFactory;
  43 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  44 import org.forester.tools.SupportCount;
  45 import org.forester.util.CommandLineArguments;
  46 import org.forester.util.ForesterUtil;
  47
  48 public class count_support {
  49
  50     final static private String  PRG_NAME                = "count_support";
  51     final static private String  PRG_VERSION             = "1.0";
  52     final static private String  PRG_DATE                = "2008.03.04";
  53     final static private String E_MAIL        = "phylosoft@gmail.com";
  54     final static private String WWW           = "www.phylosoft.org/forester/";
  55     private final static boolean WRITE_EVALUATORS_AS_NHX = false;
  56
  57     public static void main( final String args[] ) {
  58          ForesterUtil.printProgramInformation( PRG_NAME, null, PRG_VERSION, PRG_DATE, E_MAIL, WWW, ForesterUtil.getForesterLibraryInformation() );
  59
  60         if ( ( args.length < 3 ) || ( args.length > 7 ) ) {
  61             System.out.println();
  62             System.out.println( count_support.PRG_NAME + ": wrong number of arguments" );
  63             System.out.println();
  64             System.out
  65                     .println( "Usage: \"count_support [options] <file containing phylogeny to be evaluated> <file with phylogenies to be used for evaluation> <outfile> [outfile for evaluator phylogenies, "
  66                             + "always unstripped if -t=<d> option is used, otherwise strippedness is dependent on -s option]\"\n" );
  67             System.out
  68                     .println( " Options: -s strip external nodes from evaluator phylogenies not found in phylogeny to be evaluated" );
  69             System.out.println( "        : -t=<d> threshold for similarity (0.0 to 1.0)" );
  70             System.out.println( "        : -n no branch lengths in outfile for evaluator phylogenies" );
  71             System.out.println();
  72             System.exit( -1 );
  73         }
  74         CommandLineArguments cla = null;
  75         try {
  76             cla = new CommandLineArguments( args );
  77         }
  78         catch ( final Exception e ) {
  79             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  80         }
  81         final List<String> allowed_options = new ArrayList<String>();
  82         allowed_options.add( "s" );
  83         allowed_options.add( "t" );
  84         allowed_options.add( "n" );
  85         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  86         if ( dissallowed_options.length() > 0 ) {
  87             ForesterUtil.fatalError( count_support.PRG_NAME, "Unknown option(s): " + dissallowed_options );
  88         }
  89         final File phylogeny_infile = cla.getFile( 0 );
  90         final File evaluators_infile = cla.getFile( 1 );
  91         final File phylogeny_outfile = cla.getFile( 2 );
  92         File evaluators_outfile = null;
  93         boolean branch_lengths_in_ev_out = true;
  94         if ( cla.isOptionSet( "n" ) ) {
  95             branch_lengths_in_ev_out = false;
  96         }
  97         if ( cla.getNumberOfNames() == 4 ) {
  98             evaluators_outfile = cla.getFile( 3 );
  99         }
 100         else {
 101             if ( !branch_lengths_in_ev_out ) {
 102                 ForesterUtil.fatalError( count_support.PRG_NAME,
 103                                          "Cannot use -n option if no outfile for evaluators specified" );
 104             }
 105         }
 106         Phylogeny p = null;
 107         Phylogeny[] ev = null;
 108         try {
 109             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 110             final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( phylogeny_infile, true );
 111             p = factory.create( phylogeny_infile, pp )[ 0 ];
 112         }
 113         catch ( final Exception e ) {
 114             ForesterUtil.fatalError( count_support.PRG_NAME,
 115                                      "Could not read \"" + phylogeny_infile + "\" [" + e.getMessage() + "]" );
 116         }
 117         try {
 118             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 119             final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( evaluators_infile, true );
 120             ev = factory.create( evaluators_infile, pp );
 121         }
 122         catch ( final Exception e ) {
 123             ForesterUtil.fatalError( count_support.PRG_NAME,
 124                                      "Could not read \"" + evaluators_infile + "\" [" + e.getMessage() + "]" );
 125         }
 126         boolean strip = false;
 127         if ( cla.isOptionSet( "s" ) ) {
 128             strip = true;
 129         }
 130         double threshhold = -1.0;
 131         if ( cla.isOptionSet( "t" ) ) {
 132             try {
 133                 threshhold = cla.getOptionValueAsDouble( "t" );
 134             }
 135             catch ( final Exception e ) {
 136                 ForesterUtil.fatalError( count_support.PRG_NAME, "error in command line arguments: " + e.getMessage() );
 137             }
 138             if ( ( threshhold < 0 ) || ( threshhold > 1.0 ) ) {
 139                 ForesterUtil.fatalError( count_support.PRG_NAME,
 140                                          "support threshold has to be between 0.0 and 1.0 (inclusive)" );
 141             }
 142         }
 143         List<Phylogeny> evaluator_phylogenies_above_threshold = null;
 144         try {
 145             if ( threshhold >= 0 ) {
 146                 evaluator_phylogenies_above_threshold = SupportCount.count( p, ev, strip, threshhold, true );
 147                 if ( evaluator_phylogenies_above_threshold.size() < 1 ) {
 148                     ForesterUtil.fatalError( "count_support", "appears like threshold for similarity is set too high" );
 149                 }
 150             }
 151             else {
 152                 SupportCount.count( p, ev, strip, true );
 153             }
 154         }
 155         catch ( final Exception e ) {
 156             ForesterUtil.fatalError( count_support.PRG_NAME, "Failure during support counting: " + e.getMessage() );
 157         }
 158         if ( threshhold >= 0 ) {
 159             count_support.normalizeSupport( p, 100, evaluator_phylogenies_above_threshold.size() );
 160             System.out.println( evaluator_phylogenies_above_threshold.size() + " out of " + ev.length
 161                     + " evaluator phylogenies are above threshold of " + threshhold );
 162         }
 163         try {
 164             final PhylogenyWriter w = new PhylogenyWriter();
 165             w.toPhyloXML( phylogeny_outfile, p, 1 );
 166         }
 167         catch ( final IOException e ) {
 168             ForesterUtil.fatalError( count_support.PRG_NAME, "Failure to write output [" + e.getMessage() + "]" );
 169         }
 170         System.out.println();
 171         System.out.println( "Wrote phylogeny with support values to: " + phylogeny_outfile );
 172         if ( evaluators_outfile != null ) {
 173             try {
 174                 final PhylogenyWriter w = new PhylogenyWriter();
 175                 if ( evaluator_phylogenies_above_threshold != null ) {
 176                     System.out.println( "Writing " + evaluator_phylogenies_above_threshold.size()
 177                             + " evaluator phylogenies above threshold of " + threshhold + " to: " + evaluators_outfile );
 178                     if ( count_support.WRITE_EVALUATORS_AS_NHX ) {
 179                         w.toNewHampshireX( evaluator_phylogenies_above_threshold, evaluators_outfile, ";"
 180                                 + ForesterUtil.getLineSeparator() );
 181                     }
 182                     else {
 183                         w.toNewHampshire( evaluator_phylogenies_above_threshold,
 184                                           true,
 185                                           branch_lengths_in_ev_out,
 186                                           evaluators_outfile,
 187                                           ";" + ForesterUtil.getLineSeparator() );
 188                     }
 189                 }
 190                 else {
 191                     System.out.println( "Writing " + ev.length + " evaluator phylogenies to :" + evaluators_outfile );
 192                     if ( count_support.WRITE_EVALUATORS_AS_NHX ) {
 193                         w.toNewHampshireX( Arrays.asList( ev ),
 194                                            evaluators_outfile,
 195                                            ";" + ForesterUtil.getLineSeparator() );
 196                     }
 197                     else {
 198                         w.toNewHampshire( Arrays.asList( ev ), true, branch_lengths_in_ev_out, evaluators_outfile, ";"
 199                                 + ForesterUtil.getLineSeparator() );
 200                     }
 201                 }
 202             }
 203             catch ( final IOException e ) {
 204                 ForesterUtil.fatalError( count_support.PRG_NAME, "Failure to write output [" + e.getMessage() + "]" );
 205             }
 206         }
 207         System.out.println();
 208         System.out.println( "Done." );
 209         System.out.println();
 210     }
 211
 212     private static void normalizeSupport( final Phylogeny p, final double normalized_max, final int number_phylos ) {
 213         double min = Double.MAX_VALUE;
 214         double max = -Double.MAX_VALUE;
 215         double sum = 0.0;
 216         int n = 0;
 217         for( final PhylogenyNodeIterator iter = p.iteratorPostorder(); iter.hasNext(); ) {
 218             final PhylogenyNode node = iter.next();
 219             if ( !node.isRoot() && !node.isExternal() ) {
 220                 final double b = PhylogenyMethods.getConfidenceValue( node );
 221                 if ( b > max ) {
 222                     max = b;
 223                 }
 224                 if ( ( b >= 0 ) && ( b < min ) ) {
 225                     min = b;
 226                 }
 227                 sum += b;
 228                 ++n;
 229             }
 230         }
 231         double av = sum / n;
 232         System.out.println( "Max support before normalization is    : " + max );
 233         System.out.println( "Min support before normalization is    : " + min );
 234         System.out.println( "Average support before normalization is: " + av + " (=" + sum + "/" + n + ")" );
 235         System.out.println( "Normalizing so that theoretical maximum support value is: " + normalized_max );
 236         System.out.println( "Number of phylogenies used in support analysis: " + number_phylos );
 237         final double f = normalized_max / number_phylos;
 238         min = Double.MAX_VALUE;
 239         max = -Double.MAX_VALUE;
 240         sum = 0.0;
 241         n = 0;
 242         for( final PhylogenyNodeIterator iter = p.iteratorPostorder(); iter.hasNext(); ) {
 243             final PhylogenyNode node = iter.next();
 244             if ( node.isRoot() || node.isExternal() ) {
 245                 PhylogenyMethods.setBootstrapConfidence( node, Confidence.CONFIDENCE_DEFAULT_VALUE );
 246             }
 247             else {
 248                 double b = PhylogenyMethods.getConfidenceValue( node );
 249                 b = f * b;
 250                 PhylogenyMethods.setBootstrapConfidence( node, b );
 251                 if ( b > max ) {
 252                     max = b;
 253                 }
 254                 if ( ( b >= 0 ) && ( b < min ) ) {
 255                     min = b;
 256                 }
 257                 sum += b;
 258                 ++n;
 259             }
 260         }
 261         av = sum / n;
 262         System.out.println( "Max support after normalization is    : " + max );
 263         System.out.println( "Min support after normalization is    : " + min );
 264         System.out.println( "Average support after normalization is: " + av + " (=" + sum + "/" + n + ")" );
 265     }
 266 }