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

   1 // javac -cp ~/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester/java/forester.jar
   2 // ~/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester_applications/src/org/forester/applications/subtree_feature_count.java
   3 //
   4 // java -Xmx2048m -cp
   5 // /home/czmasek/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester_applications/src/:/home/czmasek/SOFTWARE_DEV/ECLIPSE_WORKSPACE/forester/java/forester.jar
   6 // org.forester.applications.subtree_feature_count
   7
   8 package org.forester.applications;
   9
  10 import java.io.File;
  11 import java.util.ArrayList;
  12 import java.util.List;
  13 import java.util.SortedSet;
  14
  15 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
  16 import org.forester.phylogeny.Phylogeny;
  17 import org.forester.phylogeny.PhylogenyMethods;
  18 import org.forester.phylogeny.PhylogenyNode;
  19 import org.forester.phylogeny.data.Accession;
  20 import org.forester.phylogeny.data.Sequence;
  21 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  22 import org.forester.phylogeny.factories.PhylogenyFactory;
  23 import org.forester.util.CommandLineArguments;
  24 import org.forester.util.ForesterUtil;
  25
  26 public class subtree_feature_count {
  27
  28     final static private String MIN_DISTANCE_TO_ROOT_OPTION = "d";
  29     final static private String E_MAIL                      = "phylosoft@gmail.com";
  30     final static private String HELP_OPTION_1               = "help";
  31     final static private String HELP_OPTION_2               = "h";
  32     final static private String PRG_DATE                    = "131120";
  33     final static private String PRG_DESC                    = "";
  34     final static private String PRG_NAME                    = "subtree_feature_count";
  35     final static private String PRG_VERSION                 = "0.90";
  36     final static private String WWW                         = "sites.google.com/site/cmzmasek/home/software/forester";
  37
  38     public static void main( final String args[] ) {
  39         try {
  40             final CommandLineArguments cla = new CommandLineArguments( args );
  41             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( args.length < 2 ) ) {
  42                 printHelp();
  43                 System.exit( 0 );
  44             }
  45             final List<String> allowed_options = new ArrayList<String>();
  46             allowed_options.add( MIN_DISTANCE_TO_ROOT_OPTION );
  47             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  48             if ( dissallowed_options.length() > 0 ) {
  49                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  50             }
  51             final double min_distance_to_root = cla.getOptionValueAsDouble( MIN_DISTANCE_TO_ROOT_OPTION );
  52             if ( min_distance_to_root <= 0 ) {
  53                 ForesterUtil.fatalError( PRG_NAME, "attempt to use min distance to root of: " + min_distance_to_root );
  54             }
  55             final File intree_file = cla.getFile( 0 );
  56             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  57             final Phylogeny phy = factory.create( intree_file, PhyloXmlParser.createPhyloXmlParserXsdValidating() )[ 0 ];
  58             execute( phy, min_distance_to_root );
  59         }
  60         catch ( final Exception e ) {
  61             e.printStackTrace();
  62             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
  63         }
  64     }
  65
  66     private final static void execute( final Phylogeny phy, final double min_distance_to_root ) {
  67         final List<List<PhylogenyNode>> ll = PhylogenyMethods.divideIntoSubTrees( phy, min_distance_to_root );
  68         for( final List<PhylogenyNode> l : ll ) {
  69             int xray = 0;
  70             int nmr = 0;
  71             int model = 0;
  72             for( final PhylogenyNode node : l ) {
  73                 if ( node.getNodeData().isHasSequence() ) {
  74                     final Sequence seq = node.getNodeData().getSequence();
  75                     final SortedSet<Accession> xrefs = seq.getCrossReferences();
  76                     if ( !ForesterUtil.isEmpty( xrefs ) ) {
  77                         for( final Accession xref : xrefs ) {
  78                             if ( xref.getSource().equalsIgnoreCase( "pdb" ) ) {
  79                                 if ( xref.getComment().equalsIgnoreCase( "x-ray" )
  80                                         || xref.getComment().equalsIgnoreCase( "xray" ) ) {
  81                                     ++xray;
  82                                 }
  83                                 if ( xref.getComment().equalsIgnoreCase( "nmr" ) ) {
  84                                     ++nmr;
  85                                 }
  86                                 if ( xref.getComment().equalsIgnoreCase( "model" ) ) {
  87                                     ++model;
  88                                 }
  89                             }
  90                         }
  91                     }
  92                 }
  93             }
  94             final int n = l.size();
  95             final double xray_p = ForesterUtil.round( ( 100.0 * xray ) / n, 1 );
  96             final double nmr_p = ForesterUtil.round( ( 100.0 * nmr ) / n, 1 );
  97             final double model_p = ForesterUtil.round( ( 100.0 * model ) / n, 1 );
  98             final StringBuilder sb = new StringBuilder();
  99             sb.append( String.valueOf( n ) );
 100             sb.append( "\t" );
 101             sb.append( String.valueOf( xray ) );
 102             sb.append( "\t" );
 103             sb.append( String.valueOf( nmr ) );
 104             sb.append( "\t" );
 105             sb.append( String.valueOf( model ) );
 106             sb.append( "\t" );
 107             sb.append( String.valueOf( xray_p ) );
 108             sb.append( "\t" );
 109             sb.append( String.valueOf( nmr_p ) );
 110             sb.append( "\t" );
 111             sb.append( String.valueOf( model_p ) );
 112             System.out.println( sb );
 113         }
 114     }
 115
 116     private static void printHelp() {
 117         ForesterUtil.printProgramInformation( PRG_NAME,
 118                                               PRG_DESC,
 119                                               PRG_VERSION,
 120                                               PRG_DATE,
 121                                               E_MAIL,
 122                                               WWW,
 123                                               ForesterUtil.getForesterLibraryInformation() );
 124         System.out.print( "Usage: " );
 125         System.out.println( PRG_NAME + " -d=<min distance to root> <intree>" );
 126         System.out.println();
 127         System.out.println();
 128     }
 129 }