forester/java/src/org/forester/application/subtree_feature_count.java

   1
   2 package org.forester.application;
   3
   4 import java.io.File;
   5 import java.util.ArrayList;
   6 import java.util.List;
   7 import java.util.SortedSet;
   8
   9 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
  10 import org.forester.phylogeny.Phylogeny;
  11 import org.forester.phylogeny.PhylogenyNode;
  12 import org.forester.phylogeny.data.Accession;
  13 import org.forester.phylogeny.data.Sequence;
  14 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  15 import org.forester.phylogeny.factories.PhylogenyFactory;
  16 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  17 import org.forester.util.CommandLineArguments;
  18 import org.forester.util.ForesterUtil;
  19
  20 public class subtree_feature_count {
  21
  22     final static private String DEPTH_OPTION  = "d";
  23     final static private String E_MAIL        = "phylosoft@gmail.com";
  24     final static private String HELP_OPTION_1 = "help";
  25     final static private String HELP_OPTION_2 = "h";
  26     final static private String PRG_DATE      = "131120";
  27     final static private String PRG_DESC      = "";
  28     final static private String PRG_NAME      = "subtree_feature_count";
  29     final static private String PRG_VERSION   = "0.90";
  30     final static private String WWW           = "sites.google.com/site/cmzmasek/home/software/forester";
  31
  32     public static void main( final String args[] ) {
  33         try {
  34             final CommandLineArguments cla = new CommandLineArguments( args );
  35             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length < 2 ) ) {
  36                 printHelp();
  37                 System.exit( 0 );
  38             }
  39             final List<String> allowed_options = new ArrayList<String>();
  40             allowed_options.add( DEPTH_OPTION );
  41             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  42             if ( dissallowed_options.length() > 0 ) {
  43                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  44             }
  45             final double depth = cla.getOptionValueAsDouble( DEPTH_OPTION );
  46             final File intree_file = cla.getFile( 0 );
  47             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  48             final Phylogeny phy = factory.create( intree_file, PhyloXmlParser.createPhyloXmlParserXsdValidating() )[ 0 ];
  49             execute( phy, depth );
  50         }
  51         catch ( final Exception e ) {
  52             e.printStackTrace();
  53             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  54         }
  55     }
  56
  57     private static StringBuilder analyzeSubtree( final PhylogenyNode n, final double depth ) {
  58         final PhylogenyNode node = moveUp( n, depth );
  59         final List<PhylogenyNode> ext_descs = node.getAllExternalDescendants();
  60         for( final PhylogenyNode ext : ext_descs ) {
  61             if ( ext.getIndicator() != 0 ) {
  62                 throw new RuntimeException( "this should not have happened" );
  63             }
  64             ext.setIndicator( ( byte ) 1 );
  65         }
  66         int xray = 0;
  67         int nmr = 0;
  68         int model = 0;
  69         boolean is_first = true;
  70         PhylogenyNode first_node = null;
  71         PhylogenyNode last_node = null;
  72         int c = 0;
  73         for( final PhylogenyNode ext : ext_descs ) {
  74             if ( is_first ) {
  75                 first_node = ext;
  76                 is_first = false;
  77             }
  78             last_node = ext;
  79             ++c;
  80             if ( ext.getNodeData().isHasSequence() ) {
  81                 final Sequence seq = ext.getNodeData().getSequence();
  82                 final SortedSet<Accession> xrefs = seq.getCrossReferences();
  83                 if ( !ForesterUtil.isEmpty( xrefs ) ) {
  84                     for( final Accession xref : xrefs ) {
  85                         if ( xref.getSource().equalsIgnoreCase( "pdb" ) ) {
  86                             if ( xref.getComment().equalsIgnoreCase( "x-ray" )
  87                                     || xref.getComment().equalsIgnoreCase( "xray" ) ) {
  88                                 ++xray;
  89                             }
  90                             if ( xref.getComment().equalsIgnoreCase( "nmr" ) ) {
  91                                 ++nmr;
  92                             }
  93                             if ( xref.getComment().equalsIgnoreCase( "model" ) ) {
  94                                 ++model;
  95                             }
  96                         }
  97                     }
  98                 }
  99             }
 100         }
 101         final StringBuilder sb = new StringBuilder();
 102         sb.append( String.valueOf( c ) );
 103         sb.append( "\t" );
 104         sb.append( first_node.getName() );
 105         sb.append( "\t" );
 106         sb.append( last_node.getName() );
 107         sb.append( "\t" );
 108         sb.append( String.valueOf( xray ) );
 109         sb.append( "\t" );
 110         sb.append( String.valueOf( nmr ) );
 111         sb.append( "\t" );
 112         sb.append( String.valueOf( model ) );
 113         return sb;
 114     }
 115
 116     private static void execute( final Phylogeny phy, final double depth ) {
 117         setAllIndicatorsToZero( phy );
 118         for( final PhylogenyNodeIterator it = phy.iteratorExternalForward(); it.hasNext(); ) {
 119             final PhylogenyNode n = it.next();
 120             if ( n.getIndicator() != 0 ) {
 121                 continue;
 122             }
 123             final StringBuilder s = analyzeSubtree( n, depth );
 124             System.out.println( s.toString() );
 125         }
 126     }
 127
 128     private static PhylogenyNode moveUp( final PhylogenyNode node, final double depth ) {
 129         PhylogenyNode n = node;
 130         PhylogenyNode prev = node;
 131         while ( depth < n.calculateDistanceToRoot() ) {
 132             prev = n;
 133             n = n.getParent();
 134         }
 135         return prev;
 136     }
 137
 138     private static void printHelp() {
 139         ForesterUtil.printProgramInformation( PRG_NAME,
 140                                               PRG_DESC,
 141                                               PRG_VERSION,
 142                                               PRG_DATE,
 143                                               E_MAIL,
 144                                               WWW,
 145                                               ForesterUtil.getForesterLibraryInformation() );
 146         System.out.println( "Usage:" );
 147         System.out.println();
 148         System.out.println( PRG_NAME + "" );
 149         System.out.println();
 150         System.out.println( " example: " );
 151         System.out.println();
 152         System.out.println();
 153         System.out.println();
 154     }
 155
 156     private static void setAllIndicatorsToZero( final Phylogeny phy ) {
 157         for( final PhylogenyNodeIterator it = phy.iteratorPostorder(); it.hasNext(); ) {
 158             it.next().setIndicator( ( byte ) 0 );
 159         }
 160     }
 161 }