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