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

   1 //
   2
   3 package org.forester.application;
   4
   5 import java.io.FileInputStream;
   6 import java.util.ArrayList;
   7 import java.util.Arrays;
   8 import java.util.HashSet;
   9 import java.util.List;
  10 import java.util.Set;
  11
  12 import org.forester.io.parsers.FastaParser;
  13 import org.forester.msa.Msa;
  14 import org.forester.sequence.BasicSequence;
  15 import org.forester.sequence.Sequence;
  16 import org.forester.util.ForesterUtil;
  17
  18 public class aa {
  19
  20     public static void main( final String args[] ) {
  21         try {
  22             System.out.println( "STARTING..." );
  23             final List<Sequence> orig = FastaParser
  24                     .parse( new FileInputStream( "C:\\Users\\zma\\Desktop\\RRMa_domains_ext_20.fasta" ) );
  25             final Msa msa = FastaParser.parseMsa( new FileInputStream( "C:\\Users\\zma\\Desktop\\test3_sorted.fasta" ) );
  26             final Set<Sequence> all_found_seqs = new HashSet<Sequence>();
  27             for( int i = 0; i < msa.getNumberOfSequences(); ++i ) {
  28                 final String id = msa.getIdentifier( i );
  29                 final String id_ = id.substring( 0, id.indexOf( "_" ) );
  30                 final String range = id.substring( id.indexOf( "[" ) + 1, id.indexOf( "]" ) );
  31                 //System.out.println( i + ": " + id + "=>" + id_ + " " + range );
  32                 if ( ForesterUtil.isEmpty( id_ ) ) {
  33                     System.out.println( "ERROR: id is empty for: " + id );
  34                     System.exit( -1 );
  35                 }
  36                 if ( ForesterUtil.isEmpty( range ) ) {
  37                     System.out.println( "ERROR: range is empty for: " + id );
  38                     System.exit( -1 );
  39                 }
  40                 int found = 0;
  41                 final List<Sequence> found_seqs = new ArrayList<Sequence>();
  42                 for( final Sequence orig_seq : orig ) {
  43                     final String orig_seq_id = orig_seq.getIdentifier();
  44                     if ( ( orig_seq_id.indexOf( id_ ) >= 0 ) && ( orig_seq_id.indexOf( "[" + range + "]" ) >= 0 ) ) {
  45                         found++;
  46                         found_seqs.add( orig_seq );
  47                     }
  48                 }
  49                 if ( found > 0 ) {
  50                     for( final Sequence found_seq : found_seqs ) {
  51                         if ( found_seq.getLength() >= 85 ) {
  52                             all_found_seqs.add( BasicSequence.createAaSequence( id, found_seq
  53                                     .getMolecularSequenceAsString() ) );
  54                         }
  55                     }
  56                     if ( found > 1 ) {
  57                         System.out.println( i + ": " + id + "=>" + id_ + " " + range );
  58                         System.out.println( "  found: " + found );
  59                         for( final Sequence found_seq : found_seqs ) {
  60                             System.out.println( found_seq.toString() );
  61                         }
  62                     }
  63                 }
  64                 else {
  65                     System.out.println( "ERROR: not found: " + id );
  66                     System.exit( -1 );
  67                 }
  68             }
  69             final String fasta_ary[] = new String[ all_found_seqs.size() ];
  70             int i = 0;
  71             for( final Sequence sequence : all_found_seqs ) {
  72                 fasta_ary[ i ] = ">" + sequence.getIdentifier() + "\n" + sequence.getMolecularSequenceAsString();
  73                 System.out.println( sequence );
  74                 i++;
  75             }
  76             Arrays.sort( fasta_ary );
  77             for( final String element : fasta_ary ) {
  78                 System.out.println( element );
  79             }
  80             System.out.println( "DONE." );
  81         }
  82         catch ( final Exception e ) {
  83             e.printStackTrace();
  84         }
  85     }
  86 }