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