forester/java/src/org/forester/application/pplacer_summary.java

   1
   2 package org.forester.application;
   3
   4 import java.io.File;
   5 import java.util.ArrayList;
   6 import java.util.Collections;
   7 import java.util.List;
   8 import java.util.regex.Matcher;
   9 import java.util.regex.Pattern;
  10
  11 import org.forester.phylogeny.Phylogeny;
  12 import org.forester.phylogeny.PhylogenyNode;
  13 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
  14 import org.forester.phylogeny.factories.PhylogenyFactory;
  15 import org.forester.util.BasicDescriptiveStatistics;
  16
  17 public class pplacer_summary {
  18
  19     public static void main( final String args[] ) {
  20         final File indir = new File( "." );
  21         final File[] list_of_files = indir.listFiles();
  22         final List<File> infiles = new ArrayList<>();
  23         for( final File file : list_of_files ) {
  24             if ( file.isFile() && file.canRead() && file.toString().endsWith( ".sing.tre" ) ) {
  25                 infiles.add( file );
  26             }
  27         }
  28         Collections.sort( infiles );
  29         final BasicDescriptiveStatistics non_unique_placements_stats = new BasicDescriptiveStatistics();
  30         final BasicDescriptiveStatistics unexpected_top_placements_stats = new BasicDescriptiveStatistics();
  31         final BasicDescriptiveStatistics unexpected_unique_top_placements_stats = new BasicDescriptiveStatistics();
  32         for( final File infile : infiles ) {
  33             Phylogeny phys[] = null;
  34             try {
  35                 final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
  36                 phys = factory.create( infile,
  37                                        org.forester.io.parsers.util.ParserUtils
  38                                                .createParserDependingOnFileType( infile, true ) );
  39             }
  40             catch ( final Exception e ) {
  41                 System.out.println( "Could not read " + infile + ":" + e );
  42                 System.exit( -1 );
  43             }
  44             final Pattern p = Pattern.compile( "_Q_#\\d+_M=(.+)" );
  45             int total_trees = 0;
  46             int total_placements = 0;
  47             int total_expected_top_placements = 0;
  48             int total_unexpected_top_placements = 0;
  49             int total_unexpected_unique_top_placements = 0;
  50             int unique_placements = 0;
  51             int non_unique_placements = 0;
  52             int m1_placements = 0;
  53             int non_m1_placements = 0;
  54             for( final Phylogeny phy : phys ) {
  55                 ++total_trees;
  56                 final List<PhylogenyNode> nodes = phy.getNodes( p );
  57                 if ( nodes.isEmpty() ) {
  58                     System.out.println();
  59                     System.out.println( "not found!" );
  60                     System.exit( -1 );
  61                 }
  62                 else {
  63                     total_placements++;
  64                     if ( nodes.size() == 1 ) {
  65                         unique_placements++;
  66                     }
  67                     else {
  68                         non_unique_placements++;
  69                     }
  70                     final PhylogenyNode n = nodes.get( 0 );
  71                     if ( n.isExternal() && !n.isRoot() ) {
  72                         final Matcher m = p.matcher( n.getName() );
  73                         if ( m.find() ) {
  74                             final double M = Double.parseDouble( m.group( 1 ) );
  75                             if ( M > 0.999999 ) {
  76                                 m1_placements++;
  77                             }
  78                             else {
  79                                 non_m1_placements++;
  80                             }
  81                         }
  82                         else {
  83                             System.out.println();
  84                             System.out.println( "no match!" );
  85                             System.exit( -1 );
  86                         }
  87                         String sib = null;
  88                         if ( n.getChildNodeIndex() == 0 ) {
  89                             sib = n.getParent().getChildNode2().getName();
  90                         }
  91                         else if ( n.getChildNodeIndex() == 1 ) {
  92                             sib = n.getParent().getChildNode1().getName();
  93                         }
  94                         else {
  95                             System.out.println();
  96                             System.out.println( "more than two children!" );
  97                             System.exit( -1 );
  98                         }
  99                         //  System.out.println( n.getName() + "->" + sib );
 100                         if ( n.getName().startsWith( sib ) ) {
 101                             total_expected_top_placements++;
 102                         }
 103                         else {
 104                             total_unexpected_top_placements++;
 105                             if ( nodes.size() == 1 ) {
 106                                 total_unexpected_unique_top_placements++;
 107                             }
 108                         }
 109                     }
 110                 }
 111             }
 112             System.out.println();
 113             System.out.println( infile.getName() );
 114             final Pattern pa = Pattern.compile( "(\\d+)-\\d+" );
 115             final Matcher m = pa.matcher( infile.getName() );
 116             if ( m.find() ) {
 117                 final int start = Integer.parseInt( m.group( 1 ) );
 118                 //      System.out.println( start + "\t"
 119                 //              + ( ( ( double ) total_unexpected_top_placements ) / total_placements ) );
 120             }
 121             System.out.println( "total trees" + "\t" + total_trees );
 122             System.out.println( "total placements" + "\t" + total_placements );
 123             System.out.println( "total expected top placements" + "\t" + total_expected_top_placements );
 124             System.out.println( "total un-expected top placements" + "\t" + total_unexpected_top_placements );
 125             System.out.println( "total un-expected unique placements" + "\t" + total_unexpected_unique_top_placements );
 126             System.out.println( "unique placements" + "\t" + unique_placements );
 127             System.out.println( "non unique placements" + "\t" + non_unique_placements );
 128             System.out.println( "m1 placements" + "\t" + m1_placements );
 129             System.out.println( "non m1 placements" + "\t" + non_m1_placements );
 130             non_unique_placements_stats.addValue( ( ( double ) non_unique_placements ) / total_placements );
 131             unexpected_top_placements_stats
 132                     .addValue( ( ( double ) total_unexpected_top_placements ) / total_placements );
 133             unexpected_unique_top_placements_stats
 134                     .addValue( ( ( double ) total_unexpected_unique_top_placements ) / total_placements );
 135         }
 136         System.out.println( "Non-unique placements: Mean\t" + non_unique_placements_stats.arithmeticMean() );
 137         System.out.println( "Non-unique placements: Min\t" + non_unique_placements_stats.getMin() );
 138         System.out.println( "Non-unique placements: Max\t" + non_unique_placements_stats.getMax() );
 139         System.out.println( "Unexpected top-placements: Mean\t" + unexpected_top_placements_stats.arithmeticMean() );
 140         System.out.println( "Unexpected top-placements: Min\t" + unexpected_top_placements_stats.getMin() );
 141         System.out.println( "Unexpected top-placements: Max\t" + unexpected_top_placements_stats.getMax() );
 142         System.out.println( "Unexpected unique placements: Mean\t"
 143                 + unexpected_unique_top_placements_stats.arithmeticMean() );
 144         System.out.println( "Unexpected unique placements: Min\t" + unexpected_unique_top_placements_stats.getMin() );
 145         System.out.println( "Unexpected unique placements: Max\t" + unexpected_unique_top_placements_stats.getMax() );
 146     }
 147 }