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: https://sites.google.com/site/cmzmasek/home/software/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 import java.util.TreeSet;
  34
  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.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.CommandLineArguments;
  43 import org.forester.util.ForesterUtil;
  44 import org.forester.ws.seqdb.SequenceDbWsTools;
  45
  46 public class gene_tree_preprocess {
  47
  48     final static private String HELP_OPTION_1           = "help";
  49     final static private String HELP_OPTION_2           = "h";
  50     final static private String PRG_NAME                = "gene_tree_preprocess";
  51     final static private String PRG_DESC                = "gene tree preprocessing for SDI analysis";
  52     final static private String PRG_VERSION             = "1.01";
  53     final static private String PRG_DATE                = "2012.06.07";
  54     final static private String E_MAIL                  = "phylosoft@gmail.com";
  55     final static private String WWW                     = "www.phylosoft.org/forester";
  56     private final static int    DEFAULT_LINES_TO_RETURN = 50;
  57
  58     public static void main( final String[] args ) {
  59         try {
  60             final CommandLineArguments cla = new CommandLineArguments( args );
  61             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length != 1 ) ) {
  62                 printHelp();
  63                 System.exit( 0 );
  64             }
  65             final File in = cla.getFile( 0 );
  66             Phylogeny phy = null;
  67             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  68             try {
  69                 phy = factory.create( in, ParserUtils.createParserDependingOnFileType( in, true ) )[ 0 ];
  70             }
  71             catch ( final IOException e ) {
  72                 ForesterUtil.fatalError( PRG_NAME,
  73                                          "failed to read phylogeny from [" + in + "]: " + e.getLocalizedMessage() );
  74             }
  75             final File outtree = new File( ForesterUtil.removeSuffix( in.toString() )
  76                     + "_preprocessed_gene_tree.phylo.xml" );
  77             final File removed_nodes = new File( ForesterUtil.removeSuffix( in.toString() ) + "_removed_nodes.txt" );
  78             final File present_species = new File( ForesterUtil.removeSuffix( in.toString() ) + "_species_present.txt" );
  79             checkForOutputFileWriteability( outtree );
  80             checkForOutputFileWriteability( removed_nodes );
  81             checkForOutputFileWriteability( present_species );
  82             if ( phy.getNumberOfExternalNodes() < 2 ) {
  83                 ForesterUtil.fatalError( PRG_NAME, "phylogeny has " + phy.getNumberOfExternalNodes()
  84                         + " external node(s), aborting" );
  85             }
  86             final SortedSet<String> not_found = SequenceDbWsTools.obtainSeqInformation( phy,
  87                                                                                         true,
  88                                                                                         false,
  89                                                                                         DEFAULT_LINES_TO_RETURN );
  90             for( final String remove_me : not_found ) {
  91                 phy.deleteSubtree( phy.getNode( remove_me ), true );
  92             }
  93             phy.clearHashIdToNodeMap();
  94             phy.externalNodesHaveChanged();
  95             if ( phy.getNumberOfExternalNodes() < 2 ) {
  96                 ForesterUtil.fatalError( PRG_NAME,
  97                                          "after removal of unresolvable external nodes, phylogeny has "
  98                                                  + phy.getNumberOfExternalNodes() + " external node(s), aborting" );
  99             }
 100             try {
 101                 final PhylogenyWriter writer = new PhylogenyWriter();
 102                 writer.toPhyloXML( phy, 0, outtree );
 103             }
 104             catch ( final IOException e ) {
 105                 ForesterUtil.fatalError( PRG_NAME, "failed to write to [" + outtree + "]: " + e.getLocalizedMessage() );
 106             }
 107             ForesterUtil.programMessage( PRG_NAME, "wrote output phylogeny to: " + outtree );
 108             final SortedSet<String> species_set = new TreeSet<String>();
 109             for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
 110                 final PhylogenyNode node = iter.next();
 111                 if ( node.getNodeData().isHasTaxonomy() ) {
 112                     final String sn = node.getNodeData().getTaxonomy().getScientificName();
 113                     if ( !ForesterUtil.isEmpty( sn ) ) {
 114                         species_set.add( sn );
 115                     }
 116                 }
 117             }
 118             try {
 119                 final BufferedWriter out = new BufferedWriter( new FileWriter( present_species ) );
 120                 for( final String species : species_set ) {
 121                     out.write( species );
 122                     out.newLine();
 123                 }
 124                 out.close();
 125             }
 126             catch ( final IOException e ) {
 127                 ForesterUtil.fatalError( PRG_NAME,
 128                                          "failed to write to [" + present_species + "]: " + e.getLocalizedMessage() );
 129             }
 130             ForesterUtil.programMessage( PRG_NAME, "wrote present species to: " + present_species );
 131             try {
 132                 final BufferedWriter out = new BufferedWriter( new FileWriter( removed_nodes ) );
 133                 for( final String remove_me : not_found ) {
 134                     out.write( remove_me );
 135                     out.newLine();
 136                 }
 137                 out.close();
 138             }
 139             catch ( final IOException e ) {
 140                 ForesterUtil.fatalError( PRG_NAME,
 141                                          "failed to write to [" + removed_nodes + "]: " + e.getLocalizedMessage() );
 142             }
 143             ForesterUtil.programMessage( PRG_NAME, "wrote removed external nodes labels to: " + removed_nodes );
 144             ForesterUtil.programMessage( PRG_NAME, "OK" );
 145         }
 146         catch ( final Exception e ) {
 147             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 148         }
 149     }
 150
 151     private static void checkForOutputFileWriteability( final File outfile ) {
 152         final String error = ForesterUtil.isWritableFile( outfile );
 153         if ( !ForesterUtil.isEmpty( error ) ) {
 154             ForesterUtil.fatalError( PRG_NAME, error );
 155         }
 156     }
 157
 158     private static void printHelp() {
 159         ForesterUtil.printProgramInformation( PRG_NAME,
 160                                               PRG_DESC,
 161                                               PRG_VERSION,
 162                                               PRG_DATE,
 163                                               E_MAIL,
 164                                               WWW,
 165                                               ForesterUtil.getForesterLibraryInformation() );
 166         System.out.print( "Usage: " );
 167         System.out.println( PRG_NAME + " <input phylogeny file>" );
 168         System.out.println();
 169     }
 170 }