X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Fmsa%2FBasicMsa.java;h=02da257664a0817e2a58a9a5a094e847a0dd8a99;hb=2320da44a171273ec44c0adcaf991687a4780b6e;hp=d913531043ced4e5ddff2baf79e65307bc8e96ad;hpb=eb38568d7687f10934bef10881474845946777dc;p=jalview.git

diff --git a/forester/java/src/org/forester/msa/BasicMsa.java b/forester/java/src/org/forester/msa/BasicMsa.java
index d913531..02da257 100644
--- a/forester/java/src/org/forester/msa/BasicMsa.java
+++ b/forester/java/src/org/forester/msa/BasicMsa.java
@@ -36,8 +36,8 @@ import java.util.Set;
 import org.forester.io.writers.SequenceWriter;
 import org.forester.io.writers.SequenceWriter.SEQ_FORMAT;
 import org.forester.sequence.BasicSequence;
-import org.forester.sequence.Sequence;
-import org.forester.sequence.Sequence.TYPE;
+import org.forester.sequence.MolecularSequence;
+import org.forester.sequence.MolecularSequence.TYPE;
 import org.forester.util.ForesterUtil;
 
 public class BasicMsa implements Msa {
@@ -65,8 +65,8 @@ public class BasicMsa implements Msa {
     }
 
     @Override
-    public List<Sequence> asSequenceList() {
-        final List<Sequence> seqs = new ArrayList<Sequence>();
+    public List<MolecularSequence> asSequenceList() {
+        final List<MolecularSequence> seqs = new ArrayList<MolecularSequence>();
         for( int i = 0; i < getNumberOfSequences(); ++i ) {
             seqs.add( getSequence( i ) );
         }
@@ -103,12 +103,12 @@ public class BasicMsa implements Msa {
     }
 
     @Override
-    public Sequence getSequence( final int row ) {
+    public MolecularSequence getSequence( final int row ) {
         return new BasicSequence( getIdentifier( row ), _data[ row ], getType() );
     }
 
     @Override
-    public Sequence getSequence( final String id ) {
+    public MolecularSequence getSequence( final String id ) {
         for( int i = 0; i < getNumberOfSequences(); ++i ) {
             if ( getIdentifier( i ).equals( id ) ) {
                 return getSequence( i );
@@ -133,7 +133,7 @@ public class BasicMsa implements Msa {
 
     @Override
     public boolean isGapAt( final int row, final int col ) {
-        return getResidueAt( row, col ) == Sequence.GAP;
+        return getResidueAt( row, col ) == MolecularSequence.GAP;
     }
 
     @Override
@@ -143,7 +143,7 @@ public class BasicMsa implements Msa {
         }
         if ( _identifiers_set.contains( id ) ) {
             throw new IllegalArgumentException( "illegal attempt to create msa with non-unique identifiers [" + id
-                    + "]" );
+                                                + "]" );
         }
         _identifiers_set.add( id );
         _identifiers[ row ] = id;
@@ -175,6 +175,9 @@ public class BasicMsa implements Msa {
             case FASTA:
                 writeToFasta( w );
                 break;
+            case NEXUS:
+                writeToNexus( w );
+                break;
             default:
                 throw new RuntimeException( "unknown format " + format );
         }
@@ -195,10 +198,39 @@ public class BasicMsa implements Msa {
         SequenceWriter.writeSeqs( asSequenceList(), w, SEQ_FORMAT.FASTA, 100 );
     }
 
+    private void writeToNexus( final Writer w ) throws IOException {
+        final int max = determineMaxIdLength() + 1;
+        w.write( "Begin Data;" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+        w.write( "   Dimensions NTax=" + getNumberOfSequences() );
+        w.write( " NChar=" + getLength() );
+        w.write( ";" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+        w.write( "   Format DataType=Protein Interleave=No gap=-;" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+        w.write( "   Matrix" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+        for( int row = 0; row < getNumberOfSequences(); ++row ) {
+            final MolecularSequence seq = getSequence( row );
+            final String s = seq.getMolecularSequenceAsString();
+            w.write( "      " );
+            w.write( ForesterUtil.pad( getIdentifier( row ).replace( ' ', '_' ), max, ' ', false ).toString() );
+            w.write( " " );
+            w.write( s );
+            w.write( ForesterUtil.LINE_SEPARATOR );
+        }
+        w.write( "   ;" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+        w.write( "End;" );
+        w.write( ForesterUtil.LINE_SEPARATOR );
+    }
+
     private void writeToPhylip( final Writer w ) throws IOException {
         final int max = determineMaxIdLength() + 1;
+        w.write( getNumberOfSequences() + " " + getLength() );
+        w.write( ForesterUtil.LINE_SEPARATOR );
         for( int row = 0; row < getNumberOfSequences(); ++row ) {
-            w.write( ForesterUtil.pad( getIdentifier( row ), max, ' ', false ).toString() );
+            w.write( ForesterUtil.pad( getIdentifier( row ).replace( ' ', '_' ), max, ' ', false ).toString() );
             for( int col = 0; col < getLength(); ++col ) {
                 w.write( getResidueAt( row, col ) );
             }
@@ -206,14 +238,14 @@ public class BasicMsa implements Msa {
         }
     }
 
-    public static Msa createInstance( final List<Sequence> seqs ) {
+    public static Msa createInstance( final List<MolecularSequence> seqs ) {
         if ( seqs.size() < 1 ) {
             throw new IllegalArgumentException( "cannot create msa from less than one sequence" );
         }
         final int length = seqs.get( 0 ).getLength();
         final BasicMsa msa = new BasicMsa( seqs.size(), length, seqs.get( 0 ).getType() );
         for( int row = 0; row < seqs.size(); ++row ) {
-            final Sequence seq = seqs.get( row );
+            final MolecularSequence seq = seqs.get( row );
             if ( seq.getLength() != length ) {
                 throw new IllegalArgumentException( "illegal attempt to build msa from sequences of unequal length ["
                         + seq.getIdentifier() + "]" );