forester/java/src/org/forester/application/gene_tree_preprocess.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2012 Christian M. Zmasek
   6 // Copyright (C) 2008-2012 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.BufferedWriter;
  29 import java.io.File;
  30 import java.io.FileWriter;
  31 import java.io.IOException;
  32 import java.util.HashSet;
  33 import java.util.Set;
  34 import java.util.SortedSet;
  35
  36 import org.forester.archaeopteryx.tools.SequenceDataRetriver;
  37 import org.forester.io.parsers.util.ParserUtils;
  38 import org.forester.io.writers.PhylogenyWriter;
  39 import org.forester.phylogeny.Phylogeny;
  40 import org.forester.phylogeny.PhylogenyMethods;
  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.util.CommandLineArguments;
  46 import org.forester.util.ForesterUtil;
  47
  48 public class gene_tree_preprocess {
  49
  50     final static private String HELP_OPTION_1 = "help";
  51     final static private String HELP_OPTION_2 = "h";
  52     final static private String PRG_NAME      = "gene_tree_preprocess";
  53     final static private String PRG_DESC      = "gene tree preprocessing for SDI analysis";
  54     final static private String PRG_VERSION   = "1.00";
  55     final static private String PRG_DATE      = "2012.06.07";
  56     final static private String E_MAIL        = "phylosoft@gmail.com";
  57     final static private String WWW           = "www.phylosoft.org/forester/";
  58
  59     public static void main( final String[] args ) {
  60         try {
  61             final CommandLineArguments cla = new CommandLineArguments( args );
  62             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length != 1 ) ) {
  63                 printHelp();
  64                 System.exit( 0 );
  65             }
  66             final File in = cla.getFile( 0 );
  67             Phylogeny phy = null;
  68             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  69             try {
  70                 phy = factory.create( in, ParserUtils.createParserDependingOnFileType( in, true ) )[ 0 ];
  71             }
  72             catch ( final IOException e ) {
  73                 ForesterUtil.fatalError( PRG_NAME,
  74                                          "failed to read phylogeny from [" + in + "]: " + e.getLocalizedMessage() );
  75             }
  76             final File outtree = new File( ForesterUtil.removeSuffix( in.toString() )
  77                     + "_preprocessed_gene_tree.phylo.xml" );
  78             final File removed_nodes = new File( ForesterUtil.removeSuffix( in.toString() ) + "_removed_nodes.txt" );
  79             final File present_species = new File( ForesterUtil.removeSuffix( in.toString() ) + "_species_present.txt" );
  80             checkForOutputFileWriteability( outtree );
  81             checkForOutputFileWriteability( removed_nodes );
  82             checkForOutputFileWriteability( present_species );
  83             if ( phy.getNumberOfExternalNodes() < 2 ) {
  84                 ForesterUtil.fatalError( PRG_NAME, "phylogeny has " + phy.getNumberOfExternalNodes()
  85                         + " external node(s), aborting" );
  86             }
  87             final SortedSet<String> not_found = SequenceDataRetriver.obtainSeqInformation( phy, true );
  88             for( final String remove_me : not_found ) {
  89                // System.out.println( " not found: " + remove_me );
  90                 PhylogenyMethods.removeNode( phy.getNode( remove_me ), phy );
  91             }
  92             if ( phy.getNumberOfExternalNodes() < 2 ) {
  93                 ForesterUtil.fatalError( PRG_NAME,
  94                                          "after removal of unresolvable external nodes, phylogeny has "
  95                                                  + phy.getNumberOfExternalNodes() + " external node(s), aborting" );
  96             }
  97             try {
  98                 final PhylogenyWriter writer = new PhylogenyWriter();
  99                 writer.toPhyloXML( phy, 0, outtree );
 100             }
 101             catch ( final IOException e ) {
 102                 ForesterUtil.fatalError( PRG_NAME, "failed to write to [" + outtree + "]: " + e.getLocalizedMessage() );
 103             }
 104             ForesterUtil.programMessage( PRG_NAME, "wrote output phylogeny to: " + outtree );
 105             final Set<String> species_found = new HashSet<String>();
 106             try {
 107                 final BufferedWriter out = new BufferedWriter( new FileWriter( present_species ) );
 108                 for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
 109                     final PhylogenyNode node = iter.next();
 110                     if ( node.getNodeData().isHasTaxonomy() ) {
 111                         final String sn = node.getNodeData().getTaxonomy().getScientificName();
 112                         if ( !ForesterUtil.isEmpty( sn ) ) {
 113                             if ( !species_found.contains( sn ) ) {
 114                                 species_found.add( sn );
 115                                 out.write( node.getNodeData().getTaxonomy().getScientificName() );
 116                                 out.newLine();
 117                             }
 118                         }
 119                     }
 120                 }
 121                 out.close();
 122             }
 123             catch ( final IOException e ) {
 124                 ForesterUtil.fatalError( PRG_NAME,
 125                                          "failed to write to [" + present_species + "]: " + e.getLocalizedMessage() );
 126             }
 127             ForesterUtil.programMessage( PRG_NAME, "wrote present species to: " + present_species );
 128             try {
 129                 final BufferedWriter out = new BufferedWriter( new FileWriter( removed_nodes ) );
 130                 for( final String remove_me : not_found ) {
 131                     out.write( remove_me );
 132                     out.newLine();
 133                 }
 134                 out.close();
 135             }
 136             catch ( final IOException e ) {
 137                 ForesterUtil.fatalError( PRG_NAME,
 138                                          "failed to write to [" + removed_nodes + "]: " + e.getLocalizedMessage() );
 139             }
 140             ForesterUtil.programMessage( PRG_NAME, "wrote removed external nodes labels to: " + removed_nodes );
 141             ForesterUtil.programMessage( PRG_NAME, "OK" );
 142         }
 143         catch ( final Exception e ) {
 144             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 145         }
 146     }
 147
 148     public static void checkForOutputFileWriteability( final File outfile ) {
 149         final String error = ForesterUtil.isWritableFile( outfile );
 150         if ( !ForesterUtil.isEmpty( error ) ) {
 151             ForesterUtil.fatalError( PRG_NAME, error );
 152         }
 153     }
 154
 155     private static void printHelp() {
 156         ForesterUtil.printProgramInformation( PRG_NAME,
 157                                               PRG_DESC,
 158                                               PRG_VERSION,
 159                                               PRG_DATE,
 160                                               E_MAIL,
 161                                               WWW,
 162                                               ForesterUtil.getForesterLibraryInformation() );
 163         System.out.print( "Usage: " );
 164         System.out.println( PRG_NAME + " <input phylogeny file>" );
 165         System.out.println();
 166     }
 167 }