forester/java/src/org/forester/application/support_statistics.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: https://sites.google.com/site/cmzmasek/home/software/forester
  25
  26 package org.forester.application;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.io.PrintWriter;
  31 import java.util.ArrayList;
  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.phylogeny.Phylogeny;
  37 import org.forester.phylogeny.PhylogenyMethods;
  38 import org.forester.phylogeny.PhylogenyNode;
  39 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  40 import org.forester.phylogeny.factories.PhylogenyFactory;
  41 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  42 import org.forester.util.BasicDescriptiveStatistics;
  43 import org.forester.util.CommandLineArguments;
  44 import org.forester.util.DescriptiveStatistics;
  45 import org.forester.util.ForesterUtil;
  46
  47 public final class support_statistics {
  48
  49     final static private int    PLACES            = 2;
  50     final static private String HELP_OPTION       = "help";
  51     final static private String OUTPUTFILE_OPTION = "o";
  52     final static private String PRG_NAME          = "support_statistics";
  53     final static private String PRG_VERSION       = "1.0";
  54     final static private String PRG_DATE          = "2008.08.29";
  55
  56     private static StringBuffer analyze( final File[] phylogenies_infiles, final Phylogeny[] phylogenies ) {
  57         final DescriptiveStatistics[] dss = new DescriptiveStatistics[ phylogenies.length ];
  58         for( int i = 0; i < phylogenies.length; i++ ) {
  59             dss[ i ] = new BasicDescriptiveStatistics();
  60             final Phylogeny p = phylogenies[ i ];
  61             for( final PhylogenyNodeIterator iter = p.iteratorPostorder(); iter.hasNext(); ) {
  62                 final PhylogenyNode node = iter.next();
  63                 if ( !node.isRoot() && !node.isExternal() ) {
  64                     double s = PhylogenyMethods.getConfidenceValue( node );
  65                     if ( s < 0.0 ) {
  66                         s = 0.0;
  67                     }
  68                     dss[ i ].addValue( s );
  69                 }
  70             }
  71         }
  72         DescriptiveStatistics dss_comp = null;
  73         if ( dss.length > 2 ) {
  74             dss_comp = new BasicDescriptiveStatistics();
  75             for( final DescriptiveStatistics element : dss ) {
  76                 dss_comp.addValue( element.arithmeticMean() );
  77             }
  78         }
  79         int max_length = 30;
  80         for( int i = 0; i < phylogenies.length; i++ ) {
  81             final int l = phylogenies_infiles[ i ].getName().length();
  82             if ( l > max_length ) {
  83                 max_length = l;
  84             }
  85         }
  86         final StringBuffer sb = new StringBuffer();
  87         sb.append( "\t" + ForesterUtil.normalizeString( "name:", max_length, true, ' ' ) + "\t" );
  88         sb.append( "median:" + "\t" );
  89         sb.append( "mean:" + "\t" );
  90         sb.append( "sd:" + "\t" );
  91         sb.append( "min:" + "\t" );
  92         sb.append( "max:" + "\t" );
  93         sb.append( "n:" + "\t" );
  94         if ( dss_comp != null ) {
  95             sb.append( "\"z-score\":" );
  96         }
  97         sb.append( ForesterUtil.getLineSeparator() );
  98         for( int i = 0; i < phylogenies.length; i++ ) {
  99             sb.append( i + 1 + ":\t"
 100                     + ForesterUtil.normalizeString( phylogenies_infiles[ i ].getName(), max_length, true, ' ' ) + "\t" );
 101             sb.append( ForesterUtil.round( dss[ i ].median(), support_statistics.PLACES ) + "\t" );
 102             sb.append( ForesterUtil.round( dss[ i ].arithmeticMean(), support_statistics.PLACES ) + "\t" );
 103             try {
 104                 sb.append( ForesterUtil.round( dss[ i ].sampleStandardDeviation(), support_statistics.PLACES ) + "\t" );
 105             }
 106             catch ( final ArithmeticException ex ) {
 107                 sb.append( "n/a\t" );
 108             }
 109             sb.append( ForesterUtil.round( dss[ i ].getMin(), support_statistics.PLACES ) + "\t" );
 110             sb.append( ForesterUtil.round( dss[ i ].getMax(), support_statistics.PLACES ) + "\t" );
 111             sb.append( dss[ i ].getN() + "\t" );
 112             if ( dss_comp != null ) {
 113                 final double z_score = dss_comp.sampleStandardUnit( dss[ i ].arithmeticMean() );
 114                 sb.append( ForesterUtil.round( z_score, support_statistics.PLACES ) + "\t" );
 115             }
 116             sb.append( ForesterUtil.getLineSeparator() );
 117         }
 118         if ( dss_comp != null ) {
 119             sb.append( ForesterUtil.getLineSeparator() );
 120             sb.append( "\t" + ForesterUtil.normalizeString( "values for support means:", max_length, true, ' ' )
 121                     + "\t\t" );
 122             sb.append( ForesterUtil.round( dss_comp.arithmeticMean(), support_statistics.PLACES ) + "\t" );
 123             sb.append( ForesterUtil.round( dss_comp.sampleStandardDeviation(), support_statistics.PLACES ) + "\t" );
 124             sb.append( ForesterUtil.round( dss_comp.getMin(), support_statistics.PLACES ) + "\t" );
 125             sb.append( ForesterUtil.round( dss_comp.getMax(), support_statistics.PLACES ) + "\t" );
 126         }
 127         return sb;
 128     }
 129
 130     public static void main( final String args[] ) {
 131         ForesterUtil.printProgramInformation( support_statistics.PRG_NAME,
 132                                               support_statistics.PRG_VERSION,
 133                                               support_statistics.PRG_DATE );
 134         if ( ( args.length < 1 ) ) {
 135             System.out.println();
 136             System.out.println( "wrong number of arguments" );
 137             System.out.println();
 138             support_statistics.printHelp();
 139             System.exit( -1 );
 140         }
 141         CommandLineArguments cla = null;
 142         try {
 143             cla = new CommandLineArguments( args );
 144         }
 145         catch ( final Exception e ) {
 146             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 147         }
 148         if ( cla.isOptionSet( support_statistics.HELP_OPTION ) ) {
 149             System.out.println();
 150             support_statistics.printHelp();
 151             System.exit( 0 );
 152         }
 153         final List<String> allowed_options = new ArrayList<String>();
 154         allowed_options.add( support_statistics.OUTPUTFILE_OPTION );
 155         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
 156         if ( dissallowed_options.length() > 0 ) {
 157             ForesterUtil.fatalError( support_statistics.PRG_NAME, "unknown option(s): " + dissallowed_options );
 158         }
 159         final File[] phylogenies_infiles = new File[ cla.getNumberOfNames() ];
 160         for( int i = 0; i < phylogenies_infiles.length; ++i ) {
 161             phylogenies_infiles[ i ] = cla.getFile( i );
 162         }
 163         File outfile = null;
 164         if ( cla.isOptionSet( support_statistics.OUTPUTFILE_OPTION ) ) {
 165             try {
 166                 outfile = new File( cla.getOptionValue( support_statistics.OUTPUTFILE_OPTION ) );
 167             }
 168             catch ( final IllegalArgumentException e ) {
 169                 ForesterUtil.fatalError( support_statistics.PRG_NAME, "error in command line: " + e.getMessage() );
 170             }
 171             final String error = ForesterUtil.isWritableFile( outfile );
 172             if ( error != null ) {
 173                 ForesterUtil.fatalError( support_statistics.PRG_NAME, error );
 174             }
 175         }
 176         final Phylogeny[] phylogenies = new Phylogeny[ phylogenies_infiles.length ];
 177         for( int i = 0; i < phylogenies_infiles.length; i++ ) {
 178             try {
 179                 final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
 180                 final PhylogenyParser pp = ParserUtils.createParserDependingOnFileType( phylogenies_infiles[ i ], true );
 181                 phylogenies[ i ] = factory.create( phylogenies_infiles[ i ], pp )[ 0 ];
 182             }
 183             catch ( final IOException e ) {
 184                 ForesterUtil.fatalError( support_statistics.PRG_NAME, "could not read \"" + phylogenies_infiles[ i ]
 185                         + "\": " + e.getMessage() );
 186             }
 187         }
 188         final StringBuffer sb = support_statistics.analyze( phylogenies_infiles, phylogenies );
 189         System.out.println();
 190         System.out.println( sb );
 191         System.out.println();
 192         if ( outfile != null ) {
 193             try {
 194                 final PrintWriter out = new PrintWriter( outfile );
 195                 out.println( sb );
 196                 out.flush();
 197                 out.close();
 198                 System.out.println( "wrote file: " + outfile );
 199             }
 200             catch ( final IOException e ) {
 201                 ForesterUtil.fatalError( support_statistics.PRG_NAME, "failed to write output: " + e.getMessage() );
 202             }
 203         }
 204         System.out.println( support_statistics.PRG_NAME + ": successfully completed" );
 205         System.out.println();
 206     }
 207
 208     private static void printHelp() {
 209         System.out.println( "usage:" );
 210         System.out.println();
 211         System.out.println( support_statistics.PRG_NAME + " [-o=<outfile>] <phylogeny infile 1> "
 212                 + "<phylogeny infile 2> <phylogeny infile 3> ..." );
 213         System.out.println();
 214         System.out.println( " options: " );
 215         System.out.println();
 216         System.out.println( " -o=<outfile> : write output to file" );
 217         System.out.println();
 218     }
 219 }