forester/java/src/org/forester/application/aa.java

   1
   2 package org.forester.application;
   3
   4 import java.io.FileInputStream;
   5 import java.util.ArrayList;
   6 import java.util.HashSet;
   7 import java.util.List;
   8 import java.util.Set;
   9 import java.util.TreeSet;
  10
  11 import org.forester.io.parsers.FastaParser;
  12 import org.forester.msa.Msa;
  13 import org.forester.sequence.Sequence;
  14 import org.forester.util.ForesterUtil;
  15
  16 public class aa {
  17
  18     public static void main( final String args[] ) {
  19         try {
  20             System.out.println( "STARTING..." );
  21             final List<Sequence> orig = FastaParser
  22                     .parse( new FileInputStream( "C:\\Users\\zma\\Desktop\\RRMa_domains_ext_20.fasta" ) );
  23             final Msa msa = FastaParser.parseMsa( new FileInputStream( "C:\\Users\\zma\\Desktop\\test3_sorted.fasta" ) );
  24             final Set<Sequence> all_found_seqs = new HashSet<Sequence>();
  25             for( int i = 0; i < msa.getNumberOfSequences(); ++i ) {
  26                 final String id = msa.getIdentifier( i );
  27                 final String id_ = id.substring( 0, id.indexOf( "_" ) );
  28                 final String range = id.substring( id.indexOf( "[" ) + 1, id.indexOf( "]" ) );
  29                 //System.out.println( i + ": " + id + "=>" + id_ + " " + range );
  30                 if ( ForesterUtil.isEmpty( id_ ) ) {
  31                     System.out.println( "ERROR: id is empty for: " + id );
  32                     System.exit( -1 );
  33                 }
  34                 if ( ForesterUtil.isEmpty( range ) ) {
  35                     System.out.println( "ERROR: range is empty for: " + id );
  36                     System.exit( -1 );
  37                 }
  38                 int found = 0;
  39                 final List<Sequence> found_seqs = new ArrayList<Sequence>();
  40                 for( final Sequence orig_seq : orig ) {
  41                     final String orig_seq_id = orig_seq.getIdentifier();
  42                     if ( orig_seq_id.indexOf( id_ ) >= 0 && orig_seq_id.indexOf( "[" + range + "]" ) >= 0 ) {
  43                         found++;
  44                         found_seqs.add( orig_seq );
  45                     }
  46                 }
  47                 if ( found > 0 ) {
  48                     for( final Sequence found_seq : found_seqs ) {
  49                         if ( found_seq.getLength() >= 85 ) {
  50                             all_found_seqs.add( found_seq );
  51                         }
  52                     }
  53                     if ( found > 1 ) {
  54                         System.out.println( i + ": " + id + "=>" + id_ + " " + range );
  55                         System.out.println( "  found: " + found );
  56                         for( final Sequence found_seq : found_seqs ) {
  57                             System.out.println( found_seq.toString() );
  58                         }
  59                     }
  60                 }
  61                 else {
  62                     System.out.println( "ERROR: not found: " + id );
  63                     System.exit( -1 );
  64                 }
  65             }
  66             int i = 0;
  67             for( Sequence sequence : all_found_seqs ) {
  68                 System.out.println( i++ + ": " + sequence );
  69             }
  70             System.out.println( "DONE." );
  71         }
  72         catch ( final Exception e ) {
  73             e.printStackTrace();
  74         }
  75     }
  76 }