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 ff6d570..1a94b74 100644 (file)
--- a/forester/java/src/org/forester/msa/MsaMethods.java
+++ b/forester/java/src/org/forester/msa/MsaMethods.java
@@ -21,7 +21,7 @@
 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
 //
 // Contact: phylosoft @ gmail . com
-// WWW: www.phylosoft.org/forester
+// WWW: https://sites.google.com/site/cmzmasek/home/software/forester
 
 package org.forester.msa;
 
@@ -65,18 +65,44 @@ public final class MsaMethods {
     public static int calcGapSumPerColumn( final Msa msa, final int col ) {
         int gap_rows = 0;
         for( int j = 0; j < msa.getNumberOfSequences(); ++j ) {
-            if ( msa.getResidueAt( j, col ) == Sequence.GAP ) {
+            if ( msa.isGapAt( j, col ) ) {
                 gap_rows++;
             }
         }
         return gap_rows;
     }
 
+    final public static Msa removeSequence( final Msa msa, final String to_remove_id ) {
+        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( msa.getSequence( row ) );
+            }
+        }
+        if ( seqs.size() < 1 ) {
+            return null;
+        }
+        return BasicMsa.createInstance( seqs );
+    }
+
     final public static Msa removeSequences( final Msa msa, final List<String> to_remove_ids ) {
         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 ) {
+            return null;
+        }
+        return BasicMsa.createInstance( seqs );
+    }
+
+    final public static Msa removeSequencesByRow( final Msa msa, final List<Integer> to_remove_rows ) {
+        final List<Sequence> seqs = new ArrayList<Sequence>();
+        for( int row = 0; row < msa.getNumberOfSequences(); ++row ) {
+            if ( !to_remove_rows.contains( row ) ) {
+                seqs.add( msa.getSequence( row ) );
             }
         }
         if ( seqs.size() < 1 ) {
@@ -85,7 +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();
@@ -96,7 +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;
             }
@@ -133,7 +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();
@@ -149,11 +192,13 @@ public final class MsaMethods {
     public static SortedMap<Character, Integer> calculateResidueDestributionPerColumn( final Msa msa, final int column ) {
         final SortedMap<Character, Integer> map = new TreeMap<Character, Integer>();
         for( final Character r : msa.getColumnAt( column ) ) {
-            if ( !map.containsKey( r ) ) {
-                map.put( r, 1 );
-            }
-            else {
-                map.put( r, map.get( r ) + 1 );
+            if ( r != Sequence.GAP ) {
+                if ( !map.containsKey( r ) ) {
+                    map.put( r, 1 );
+                }
+                else {
+                    map.put( r, map.get( r ) + 1 );
+                }
             }
         }
         return map;
@@ -166,4 +211,32 @@ public final class MsaMethods {
         }
         return stats;
     }
+
+    public static Msa removeSequencesByMinimalLength( final Msa msa, final int min_effective_length ) {
+        final List<Integer> to_remove_rows = new ArrayList<Integer>();
+        for( int seq = 0; seq < msa.getNumberOfSequences(); ++seq ) {
+            int eff_length = 0;
+            for( int i = 0; i < msa.getLength(); ++i ) {
+                if ( msa.getResidueAt( seq, i ) != Sequence.GAP ) {
+                    eff_length++;
+                }
+            }
+            if ( eff_length < min_effective_length ) {
+                to_remove_rows.add( seq );
+            }
+        }
+        return removeSequencesByRow( msa, to_remove_rows );
+    }
+
+    public static double calcGapRatio( final Msa msa ) {
+        int gaps = 0;
+        for( int seq = 0; seq < msa.getNumberOfSequences(); ++seq ) {
+            for( int i = 0; i < msa.getLength(); ++i ) {
+                if ( msa.getResidueAt( seq, i ) == Sequence.GAP ) {
+                    gaps++;
+                }
+            }
+        }
+        return ( double ) gaps / ( msa.getLength() * msa.getNumberOfSequences() );
+    }
 }