forester_applications/src/org/forester/applications/get_subtree_specific_chars.java

   1
   2 package org.forester.applications;
   3
   4 // $Id:
   5 // FORESTER -- software libraries and applications
   6 // for evolutionary biology research and applications.
   7 //
   8 // Copyright (C) 2008-2011 Christian M. Zmasek
   9 // Copyright (C) 2008-2011 Burnham Institute for Medical Research
  10 // All rights reserved
  11 //
  12 // This library is free software; you can redistribute it and/or
  13 // modify it under the terms of the GNU Lesser General Public
  14 // License as published by the Free Software Foundation; either
  15 // version 2.1 of the License, or (at your option) any later version.
  16 //
  17 // This library is distributed in the hope that it will be useful,
  18 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  20 // Lesser General Public License for more details.
  21 //
  22 // You should have received a copy of the GNU Lesser General Public
  23 // License along with this library; if not, write to the Free Software
  24 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  25 //
  26 // Contact: phylosoft @ gmail . com
  27 // WWW: www.phylosoft.org/forester
  28 // javac -cp ~/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester/java/forester.jar
  29 // ~/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester_applications/src/org/forester/applications/get_subtree_specific_chars.java
  30 // java -Xmx2048m -cp
  31 // /home/czmasek/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester_applications/src/:/home/czmasek/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester/java/forester.jar
  32 // org.forester.applications.get_subtree_specific_chars
  33 import java.io.File;
  34 import java.util.List;
  35 import java.util.SortedSet;
  36 import java.util.TreeSet;
  37
  38 import org.forester.phylogeny.Phylogeny;
  39 import org.forester.phylogeny.PhylogenyNode;
  40 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  41 import org.forester.phylogeny.factories.PhylogenyFactory;
  42 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  43 import org.forester.util.ForesterUtil;
  44
  45 public class get_subtree_specific_chars {
  46
  47     final static boolean SIMPLE = true;
  48
  49     public static void main( final String args[] ) {
  50         if ( args.length != 1 ) {
  51             System.err.println();
  52             System.err.println( "get_subtree_specific_chars: wrong number of arguments" );
  53             System.err.println( "Usage: \"get_subtree_specific_chars <intree>" );
  54             System.err.println();
  55             System.exit( -1 );
  56         }
  57         final File infile = new File( args[ 0 ] );
  58         Phylogeny phy = null;
  59         try {
  60             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  61             phy = factory.create( infile, org.forester.io.parsers.util.ParserUtils
  62                                   .createParserDependingOnFileType( infile, true ) )[ 0 ];
  63         }
  64         catch ( final Exception e ) {
  65             System.err.println( e + "\nCould not read " + infile + "\n" );
  66             System.exit( -1 );
  67         }
  68         final SortedSet<Long> all_external_ids = getAllExternalDescendantsNodeIds( phy.getRoot() );
  69         final SortedSet<String> all_chars = getAllExternalPresentAndGainedCharacters( phy.getRoot() );
  70         System.out.println( "Sum of all external characters:\t" + all_chars.size() );
  71         System.out.println();
  72         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
  73             final PhylogenyNode node = iter.next();
  74             if ( !SIMPLE && node.isExternal() ) {
  75                 continue;
  76             }
  77             if ( !node.isRoot() ) {
  78                 // System.out.println();
  79                 if ( node.getNodeData().isHasTaxonomy()
  80                         && !ForesterUtil.isEmpty( node.getNodeData().getTaxonomy().getScientificName() ) ) {
  81                     System.out.print( node.getNodeData().getTaxonomy().getScientificName() );
  82                 }
  83                 else {
  84                     System.out.print( node.getName() );
  85                 }
  86                 // System.out.println( ":" );
  87                 System.out.print( "\t" );
  88                 final SortedSet<Long> external_ids = getAllExternalDescendantsNodeIds( node );
  89                 final SortedSet<Long> not_external_ids = copy( all_external_ids );
  90                 not_external_ids.removeAll( external_ids );
  91                 final SortedSet<String> not_node_chars = new TreeSet<String>();
  92                 for( final Long id : not_external_ids ) {
  93                     not_node_chars.addAll( getAllExternalPresentAndGainedCharacters( phy.getNode( id ) ) );
  94                 }
  95                 final SortedSet<String> node_chars = getAllExternalPresentAndGainedCharacters( node );
  96                 final SortedSet<String> unique_chars = new TreeSet<String>();
  97                 for( final String node_char : node_chars ) {
  98                     if ( !not_node_chars.contains( node_char ) ) {
  99                         if ( SIMPLE ) {
 100                             unique_chars.add( node_char );
 101                         }
 102                         else {
 103                             boolean found = true;
 104                             for( final Long external_id : external_ids ) {
 105                                 if ( !phy.getNode( external_id ).getNodeData().getBinaryCharacters()
 106                                         .getGainedCharacters().contains( node_char )
 107                                         && !phy.getNode( external_id ).getNodeData().getBinaryCharacters()
 108                                         .getPresentCharacters().contains( node_char ) ) {
 109                                     found = false;
 110                                     break;
 111                                 }
 112                             }
 113                             if ( found ) {
 114                                 unique_chars.add( node_char );
 115                             }
 116                         }
 117                     }
 118                 }
 119                 // System.out.println( "\tSUM:\t" + unique_chars.size() );
 120                 // System.out.println( unique_chars.size() );
 121                 int counter = 0;
 122                 System.out.print( "\t" + unique_chars.size() );
 123                 for( final String unique_char : unique_chars ) {
 124                     // System.out.println( "\t" + counter + ":\t" + unique_char
 125                     // );
 126                     // System.out.println( "\t" + counter + ":\t" + unique_char
 127                     // );
 128                     System.out.print( "\t" + unique_char );
 129                     ++counter;
 130                 }
 131                 System.out.println();
 132             }
 133         }
 134     }
 135
 136     private static SortedSet<Long> copy( final SortedSet<Long> set ) {
 137         final SortedSet<Long> copy = new TreeSet<Long>();
 138         for( final Long i : set ) {
 139             copy.add( i );
 140         }
 141         return copy;
 142     }
 143
 144     private static SortedSet<Long> getAllExternalDescendantsNodeIds( final PhylogenyNode node ) {
 145         final SortedSet<Long> ids = new TreeSet<Long>();
 146         final List<PhylogenyNode> descs = node.getAllExternalDescendants();
 147         for( final PhylogenyNode desc : descs ) {
 148             ids.add( desc.getId() );
 149         }
 150         return ids;
 151     }
 152
 153     private static SortedSet<String> getAllExternalPresentAndGainedCharacters( final PhylogenyNode node ) {
 154         final SortedSet<String> chars = new TreeSet<String>();
 155         final List<PhylogenyNode> descs = node.getAllExternalDescendants();
 156         for( final PhylogenyNode desc : descs ) {
 157             chars.addAll( desc.getNodeData().getBinaryCharacters().getGainedCharacters() );
 158             chars.addAll( desc.getNodeData().getBinaryCharacters().getPresentCharacters() );
 159         }
 160         return chars;
 161     }
 162 }