rio - gsdir work...

[jalview.git] / forester / java / src / org / forester / msa / BasicMsa.java

diff --git a/forester/java/src/org/forester/msa/BasicMsa.java b/forester/java/src/org/forester/msa/BasicMsa.java
index fa6e321..907d212 100644 (file)
--- a/forester/java/src/org/forester/msa/BasicMsa.java
+++ b/forester/java/src/org/forester/msa/BasicMsa.java
@@ -27,8 +27,14 @@ package org.forester.msa;
 
 import java.io.IOException;
 import java.io.Writer;
+import java.util.ArrayList;
+import java.util.HashSet;
 import java.util.List;
+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.util.ForesterUtil;
@@ -36,7 +42,7 @@ import org.forester.util.ForesterUtil;
 public class BasicMsa implements Msa {
 
     private final char[][] _data;
-    private final Object[] _identifiers;
+    private final String[] _identifiers;
     private final TYPE     _type;
 
     public BasicMsa( final int rows, final int columns, final TYPE type ) {
@@ -44,7 +50,7 @@ public class BasicMsa implements Msa {
             throw new IllegalArgumentException( "basic msa of size zero are illegal" );
         }
         _data = new char[ rows ][ columns ];
-        _identifiers = new Object[ rows ];
+        _identifiers = new String[ rows ];
         _type = type;
     }
 
@@ -54,6 +60,15 @@ public class BasicMsa implements Msa {
         _type = msa._type;
     }
 
+    @Override
+    public List<Sequence> asSequenceList() {
+        final List<Sequence> seqs = new ArrayList<Sequence>();
+        for( int i = 0; i < getNumberOfSequences(); ++i ) {
+            seqs.add( getSequence( i ) );
+        }
+        return seqs;
+    }
+
     private int determineMaxIdLength() {
         int max = 0;
         for( int row = 0; row < _data.length; ++row ) {
@@ -66,7 +81,7 @@ public class BasicMsa implements Msa {
     }
 
     @Override
-    public Object getIdentifier( final int row ) {
+    public String getIdentifier( final int row ) {
         return _identifiers[ row ];
     }
 
@@ -86,6 +101,21 @@ public class BasicMsa implements Msa {
     }
 
     @Override
+    public Sequence getSequence( final String id ) {
+        for( int i = 0; i < getNumberOfSequences(); ++i ) {
+            if ( getIdentifier( i ).equals( id ) ) {
+                return getSequence( i );
+            }
+        }
+        return null;
+    }
+
+    @Override
+    public Sequence getSequence( final int row ) {
+        return new BasicSequence( getIdentifier( row ), _data[ row ], getType() );
+    }
+
+    @Override
     public StringBuffer getSequenceAsString( final int row ) {
         final StringBuffer sb = new StringBuffer( _data[ 0 ].length );
         for( int col = 0; col < _data[ 0 ].length; ++col ) {
@@ -100,7 +130,7 @@ public class BasicMsa implements Msa {
     }
 
     @Override
-    public void setIdentifier( final int row, final Object id ) {
+    public void setIdentifier( final int row, final String id ) {
         _identifiers[ row ] = id;
     }
 
@@ -124,7 +154,24 @@ public class BasicMsa implements Msa {
     }
 
     @Override
-    public void write( final Writer w ) throws IOException {
+    public void write( final Writer w, final MSA_FORMAT format ) throws IOException {
+        switch ( format ) {
+            case PHYLIP:
+                writeToPhylip( w );
+                break;
+            case FASTA:
+                writeToFasta( w );
+                break;
+            default:
+                throw new RuntimeException( "unknown format " + format );
+        }
+    }
+
+    private void writeToFasta( final Writer w ) throws IOException {
+        SequenceWriter.writeSeqs( asSequenceList(), w, SEQ_FORMAT.FASTA, 100 );
+    }
+
+    private void writeToPhylip( final Writer w ) throws IOException {
         final int max = determineMaxIdLength() + 1;
         for( int row = 0; row < _data.length; ++row ) {
             w.write( ForesterUtil.pad( _identifiers[ row ].toString(), max, ' ', false ).toString() );
@@ -139,6 +186,7 @@ public class BasicMsa implements Msa {
         if ( seqs.size() < 1 ) {
             throw new IllegalArgumentException( "cannot create basic msa from less than one sequence" );
         }
+        final Set<String> ids = new HashSet<String>();
         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 ) {
@@ -149,6 +197,11 @@ public class BasicMsa implements Msa {
             if ( seq.getType() != msa.getType() ) {
                 throw new IllegalArgumentException( "illegal attempt to build msa from sequences of different type" );
             }
+            if ( ids.contains( seq.getIdentifier() ) ) {
+                throw new IllegalArgumentException( "illegal attempt to create msa with non-unique identifiers ["
+                        + seq.getIdentifier() + "]" );
+            }
+            ids.add( seq.getIdentifier() );
             msa.setIdentifier( row, seq.getIdentifier() );
             for( int col = 0; col < length; ++col ) {
                 msa._data[ row ][ col ] = seq.getResidueAt( col );
@@ -156,4 +209,13 @@ public class BasicMsa implements Msa {
         }
         return msa;
     }
+
+    @Override
+    public List<Character> getColumnAt( final int col ) {
+        final List<Character> column = new ArrayList<Character>();
+        for( int row = 0; row < getNumberOfSequences(); ++row ) {
+            column.add( getResidueAt( row, col ) );
+        }
+        return column;
+    }
 }