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.SortedSet;
  33
  34 import org.forester.archaeopteryx.tools.SequenceDataRetriver;
  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.factories.ParserBasedPhylogenyFactory;
  40 import org.forester.phylogeny.factories.PhylogenyFactory;
  41 import org.forester.util.CommandLineArguments;
  42 import org.forester.util.ForesterUtil;
  43
  44 public class gene_tree_preprocess {
  45
  46     final static private String HELP_OPTION_1 = "help";
  47     final static private String HELP_OPTION_2 = "h";
  48     final static private String PRG_NAME      = "gene_tree_preprocess";
  49     final static private String PRG_DESC      = "gene tree preprocessing for SDI analysis";
  50     final static private String PRG_VERSION   = "1.00";
  51     final static private String PRG_DATE      = "2012.06.07";
  52     final static private String E_MAIL        = "phylosoft@gmail.com";
  53     final static private String WWW           = "www.phylosoft.org/forester/";
  54
  55     public static void main( final String[] args ) {
  56         try {
  57             final CommandLineArguments cla = new CommandLineArguments( args );
  58             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length != 1 ) ) {
  59                 printHelp();
  60                 System.exit( 0 );
  61             }
  62             final File in = cla.getFile( 0 );
  63             Phylogeny phy = null;
  64             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  65             try {
  66                 phy = factory.create( in, ParserUtils.createParserDependingOnFileType( in, true ) )[ 0 ];
  67             }
  68             catch ( final IOException e ) {
  69                 ForesterUtil.fatalError( PRG_NAME,
  70                                          "failed to read target phylogenies from [" + in + "]: "
  71                                                  + e.getLocalizedMessage() );
  72             }
  73             final File outtree = new File( ForesterUtil.removeSuffix( in.toString() )
  74                     + "_preprocessed_gene_tree.phylo.xml" );
  75             final File removed_nodes = new File( ForesterUtil.removeSuffix( in.toString() ) + "_removed_nodes.txt" );
  76             checkForOutputFileWriteability( outtree );
  77             checkForOutputFileWriteability( removed_nodes );
  78             if ( phy.getNumberOfExternalNodes() < 2 ) {
  79                 ForesterUtil.fatalError( PRG_NAME, "phylogeny has " + phy.getNumberOfExternalNodes()
  80                         + " external node(s), aborting" );
  81             }
  82             final SortedSet<String> not_found = SequenceDataRetriver.obtainSeqInformation( phy, true );
  83             for( final String remove_me : not_found ) {
  84                 System.out.println( " not found: " + remove_me );
  85                 PhylogenyMethods.removeNode( phy.getNode( remove_me ), phy );
  86             }
  87             if ( phy.getNumberOfExternalNodes() < 2 ) {
  88                 ForesterUtil.fatalError( PRG_NAME,
  89                                          "after removal of unresolvable external nodes, phylogeny has "
  90                                                  + phy.getNumberOfExternalNodes() + " external node(s), aborting" );
  91             }
  92             try {
  93                 final PhylogenyWriter writer = new PhylogenyWriter();
  94                 writer.toPhyloXML( phy, 0, outtree );
  95             }
  96             catch ( final IOException e ) {
  97                 ForesterUtil.fatalError( PRG_NAME, "failed to write to [" + outtree + "]: " + e.getLocalizedMessage() );
  98             }
  99             ForesterUtil.programMessage( PRG_NAME, "wrote output phylogeny to: " + outtree );
 100             try {
 101                 final BufferedWriter out = new BufferedWriter( new FileWriter( removed_nodes ) );
 102                 for( final String remove_me : not_found ) {
 103                     out.write( remove_me );
 104                     out.newLine();
 105                 }
 106                 out.close();
 107             }
 108             catch ( final IOException e ) {
 109                 ForesterUtil.fatalError( PRG_NAME,
 110                                          "failed to write to [" + removed_nodes + "]: " + e.getLocalizedMessage() );
 111             }
 112             ForesterUtil.programMessage( PRG_NAME, "wrote removed external nodes labels to: " + removed_nodes );
 113             ForesterUtil.programMessage( PRG_NAME, "OK" );
 114         }
 115         catch ( final Exception e ) {
 116             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 117         }
 118     }
 119
 120     public static void checkForOutputFileWriteability( final File outfile ) {
 121         final String error = ForesterUtil.isWritableFile( outfile );
 122         if ( !ForesterUtil.isEmpty( error ) ) {
 123             ForesterUtil.fatalError( PRG_NAME, error );
 124         }
 125     }
 126
 127     private static void printHelp() {
 128         ForesterUtil.printProgramInformation( PRG_NAME,
 129                                               PRG_DESC,
 130                                               PRG_VERSION,
 131                                               PRG_DATE,
 132                                               E_MAIL,
 133                                               WWW,
 134                                               ForesterUtil.getForesterLibraryInformation() );
 135         System.out.print( "Usage: " );
 136         System.out.println( PRG_NAME + " <input phylogeny file>" );
 137         System.out.println();
 138     }
 139 }