work for rrm project (ComPhy 2012 Moscow)

diff --git a/forester/java/src/org/forester/application/aa.java b/forester/java/src/org/forester/application/aa.java
index b9e88a7..6d7be4f 100644 (file)
--- a/forester/java/src/org/forester/application/aa.java
+++ b/forester/java/src/org/forester/application/aa.java
@@ -3,7 +3,10 @@ package org.forester.application;
 
 import java.io.FileInputStream;
 import java.util.ArrayList;
 
 import java.io.FileInputStream;
 import java.util.ArrayList;

+import java.util.HashSet;

 import java.util.List;
 import java.util.List;

+import java.util.Set;
+import java.util.TreeSet;

 
 import org.forester.io.parsers.FastaParser;
 import org.forester.msa.Msa;
 
 import org.forester.io.parsers.FastaParser;
 import org.forester.msa.Msa;

@@ -18,6 +21,7 @@ public class aa {
             final List<Sequence> orig = FastaParser
                     .parse( new FileInputStream( "C:\\Users\\zma\\Desktop\\RRMa_domains_ext_20.fasta" ) );
             final Msa msa = FastaParser.parseMsa( new FileInputStream( "C:\\Users\\zma\\Desktop\\test3_sorted.fasta" ) );
             final List<Sequence> orig = FastaParser
                     .parse( new FileInputStream( "C:\\Users\\zma\\Desktop\\RRMa_domains_ext_20.fasta" ) );
             final Msa msa = FastaParser.parseMsa( new FileInputStream( "C:\\Users\\zma\\Desktop\\test3_sorted.fasta" ) );

+            final Set<Sequence> all_found_seqs = new HashSet<Sequence>();

             for( int i = 0; i < msa.getNumberOfSequences(); ++i ) {
                 final String id = msa.getIdentifier( i );
                 final String id_ = id.substring( 0, id.indexOf( "_" ) );
             for( int i = 0; i < msa.getNumberOfSequences(); ++i ) {
                 final String id = msa.getIdentifier( i );
                 final String id_ = id.substring( 0, id.indexOf( "_" ) );

diff --git a/forester/java/src/org/forester/sequence/BasicSequence.java b/forester/java/src/org/forester/sequence/BasicSequence.java
index c14277c..e180481 100644 (file)
--- a/forester/java/src/org/forester/sequence/BasicSequence.java
+++ b/forester/java/src/org/forester/sequence/BasicSequence.java
@@ -26,6 +26,8 @@
 
 package org.forester.sequence;
 
 
 package org.forester.sequence;
 

+import org.forester.util.ForesterUtil;
+

 public class BasicSequence implements Sequence {
 
     private final char[] _mol_sequence;
 public class BasicSequence implements Sequence {
 
     private final char[] _mol_sequence;

@@ -33,6 +35,12 @@ public class BasicSequence implements Sequence {
     private final TYPE   _type;
 
     private BasicSequence( final String identifier, final String mol_sequence, final TYPE type ) {
     private final TYPE   _type;
 
     private BasicSequence( final String identifier, final String mol_sequence, final TYPE type ) {

+        if ( ForesterUtil.isEmpty( identifier ) ) {
+            throw new IllegalArgumentException( "identifier of sequence cannot be empty");
+        }
+        if ( ForesterUtil.isEmpty( mol_sequence ) ) {
+            throw new IllegalArgumentException( "molecular sequence cannot be empty");
+        }

         _mol_sequence = mol_sequence.toCharArray();
         _identifier = identifier;
         _type = type;
         _mol_sequence = mol_sequence.toCharArray();
         _identifier = identifier;
         _type = type;

@@ -40,6 +48,12 @@ public class BasicSequence implements Sequence {
 
     // Only use if you know what you are doing!
     public BasicSequence( final String identifier, final char[] mol_sequence, final TYPE type ) {
 
     // Only use if you know what you are doing!
     public BasicSequence( final String identifier, final char[] mol_sequence, final TYPE type ) {

+        if ( ForesterUtil.isEmpty( identifier ) ) {
+            throw new IllegalArgumentException( "identifier of sequence cannot be empty");
+        }
+        if ( mol_sequence == null || mol_sequence.length < 1 ) {
+            throw new IllegalArgumentException( "molecular sequence cannot be empty");
+        }

         _mol_sequence = mol_sequence;
         _identifier = identifier;
         _type = type;
         _mol_sequence = mol_sequence;
         _identifier = identifier;
         _type = type;

@@ -82,11 +96,32 @@ public class BasicSequence implements Sequence {
     }
 
     @Override
     }
 
     @Override

+    public boolean equals(Object obj) {
+        if (obj == null) {
+            return false;
+        }
+        if (obj.getClass() != getClass()) {
+            return false;
+        }
+        Sequence other = ( Sequence) obj;
+        if ( getMolecularSequenceAsString() .equals( other.getMolecularSequenceAsString() )) {
+            return true;
+        }
+        return false;
+    }
+ 
+    @Override
+    public int hashCode() {
+        return getMolecularSequenceAsString().hashCode();
+    }
+    
+    
+    @Override

     public String toString() {
         final StringBuffer sb = new StringBuffer();
         sb.append( _identifier.toString() );
     public String toString() {
         final StringBuffer sb = new StringBuffer();
         sb.append( _identifier.toString() );

-        sb.append( " " );
-        sb.append( new String( _mol_sequence ) );
+        sb.append( ": " );
+        sb.append( getMolecularSequenceAsString() );

         return sb.toString();
     }
 
         return sb.toString();
     }
 

@@ -112,4 +147,10 @@ public class BasicSequence implements Sequence {
         return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
                 .replaceAll( RNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.RNA );
     }
         return new BasicSequence( identifier, mol_sequence.toUpperCase().replaceAll( "\\.", GAP_STR )
                 .replaceAll( RNA_REGEXP, Character.toString( UNSPECIFIED_NUC ) ), TYPE.RNA );
     }

+
+    @Override
+    public String getMolecularSequenceAsString() {
+       
+        return new String( getMolecularSequence() );
+    }

 }
 }

diff --git a/forester/java/src/org/forester/sequence/Sequence.java b/forester/java/src/org/forester/sequence/Sequence.java
index 9a92fbe..f4c17cf 100644 (file)
--- a/forester/java/src/org/forester/sequence/Sequence.java
+++ b/forester/java/src/org/forester/sequence/Sequence.java
@@ -44,6 +44,8 @@ public interface Sequence {
     public abstract int getNumberOfGapResidues();
 
     public abstract char[] getMolecularSequence();
     public abstract int getNumberOfGapResidues();
 
     public abstract char[] getMolecularSequence();

+    
+    public abstract String getMolecularSequenceAsString();

 
     public abstract char getResidueAt( final int position );
 
 
     public abstract char getResidueAt( final int position );