inprogress

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

diff --git a/forester/java/src/org/forester/msa/MsaMethods.java b/forester/java/src/org/forester/msa/MsaMethods.java
index ac6cb1f..1a94b74 100644 (file)
--- a/forester/java/src/org/forester/msa/MsaMethods.java
+++ b/forester/java/src/org/forester/msa/MsaMethods.java
@@ -76,7 +76,7 @@ public final class MsaMethods {
         final List<Sequence> seqs = new ArrayList<Sequence>();
         for( int row = 0; row < msa.getNumberOfSequences(); ++row ) {
             if ( !to_remove_id.equals( msa.getIdentifier( row ) ) ) {
-                seqs.add( BasicSequence.copySequence( msa.getSequence( row ) ) );
+                seqs.add( msa.getSequence( row ) );
             }
         }
         if ( seqs.size() < 1 ) {
@@ -89,7 +89,7 @@ public final class MsaMethods {
         final List<Sequence> seqs = new ArrayList<Sequence>();
         for( int row = 0; row < msa.getNumberOfSequences(); ++row ) {
             if ( !to_remove_ids.contains( msa.getIdentifier( row ) ) ) {
-                seqs.add( BasicSequence.copySequence( msa.getSequence( row ) ) );
+                seqs.add( msa.getSequence( row ) );
             }
         }
         if ( seqs.size() < 1 ) {
@@ -102,7 +102,7 @@ public final class MsaMethods {
         final List<Sequence> seqs = new ArrayList<Sequence>();
         for( int row = 0; row < msa.getNumberOfSequences(); ++row ) {
             if ( !to_remove_rows.contains( row ) ) {
-                seqs.add( BasicSequence.copySequence( msa.getSequence( row ) ) );
+                seqs.add( msa.getSequence( row ) );
             }
         }
         if ( seqs.size() < 1 ) {
@@ -111,7 +111,7 @@ public final class MsaMethods {
         return BasicMsa.createInstance( seqs );
     }
 
-    synchronized final public Msa removeGapColumns( final double max_allowed_gap_ratio,
+    synchronized final public Msa deleteGapColumns( final double max_allowed_gap_ratio,
                                                     final int min_allowed_length,
                                                     final Msa msa ) {
         init();
@@ -122,7 +122,7 @@ public final class MsaMethods {
         final boolean[] delete_cols = new boolean[ msa.getLength() ];
         int new_length = 0;
         for( int col = 0; col < msa.getLength(); ++col ) {
-            delete_cols[ col ] = ( ( double ) calcGapSumPerColumn( msa, col ) / msa.getNumberOfSequences() ) >= max_allowed_gap_ratio;
+            delete_cols[ col ] = ( ( double ) calcGapSumPerColumn( msa, col ) / msa.getNumberOfSequences() ) > max_allowed_gap_ratio;
             if ( !delete_cols[ col ] ) {
                 ++new_length;
             }
@@ -159,7 +159,24 @@ public final class MsaMethods {
         return BasicMsa.createInstance( seqs );
     }
 
-    public static double calculateIdentityRatio( final Msa msa, final int column ) {
+    final public static DescriptiveStatistics calculateIdentityRatio( final int from, final int to, final Msa msa ) {
+        final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
+        for( int c = from; c <= to; ++c ) {
+            stats.addValue( calculateIdentityRatio( msa, c ) );
+        }
+        return stats;
+    }
+
+    final public static DescriptiveStatistics calculateEffectiveLengthStatistics( final Msa msa ) {
+        final DescriptiveStatistics stats = new BasicDescriptiveStatistics();
+        for( int row = 0; row < msa.getNumberOfSequences(); ++row ) {
+            final Sequence s = msa.getSequence( row );
+            stats.addValue( s.getLength() - s.getNumberOfGapResidues() );
+        }
+        return stats;
+    }
+
+    final public static double calculateIdentityRatio( final Msa msa, final int column ) {
         final SortedMap<Character, Integer> dist = calculateResidueDestributionPerColumn( msa, column );
         int majority_count = 0;
         final Iterator<Map.Entry<Character, Integer>> it = dist.entrySet().iterator();