forester/java/src/org/forester/msa/MsaCompactor.java

   1 package org.forester.msa;
   2
   3 import org.forester.sequence.Sequence;
   4 import org.forester.util.BasicDescriptiveStatistics;
   5 import org.forester.util.DescriptiveStatistics;
   6
   7
   8 public class MsaCompactor {
   9
  10     final private Msa _msa;
  11
  12     public MsaCompactor( Msa msa ) {
  13
  14         _msa = msa;
  15     }
  16
  17
  18
  19     private DescriptiveStatistics[] calc() {
  20         final double gappiness[] = calcGappiness();
  21         final DescriptiveStatistics stats[] = new DescriptiveStatistics[ _msa.getNumberOfSequences() ];
  22         for ( int row = 0; row < _msa.getNumberOfSequences(); ++row ) {
  23             stats[ row ] = new BasicDescriptiveStatistics();
  24             for( int col = 0; col < _msa.getLength(); ++col ) {
  25                 if ( _msa.getResidueAt( row, col ) != Sequence.GAP ) {
  26                     stats[ row ].addValue( gappiness[ col ] );
  27
  28                 }
  29             }
  30         }
  31         return stats;
  32     }
  33
  34     private double[] calcGappiness() {
  35         final double gappiness[] = new double[ _msa.getLength() ];
  36         final int seqs =  _msa.getNumberOfSequences();
  37         for( int i = 0; i < gappiness.length; ++i ) {
  38             gappiness[ i ] = ( double ) MsaMethods.calcGapSumPerColumn( _msa, i )  / _msa.getNumberOfSequences();
  39
  40         }
  41         return gappiness;
  42
  43     }
  44
  45 }