X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Fmsa%2FMsaMethods.java;h=06ca50bce885d4f83c59eb13b364702ed17ef540;hb=f3f08cb017c276411ba6c7e3f3e157ed5bba136e;hp=716c8431211cffb62118b3f183c7667bb61ac36d;hpb=7476d4a39fc5840955b57289b862bf08b8419fd7;p=jalview.git

diff --git a/forester/java/src/org/forester/msa/MsaMethods.java b/forester/java/src/org/forester/msa/MsaMethods.java
index 716c843..06ca50b 100644
--- 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;
 
@@ -72,9 +72,35 @@ public final class MsaMethods {
         return gap_rows;
     }
 
-    synchronized public Msa removeGapColumns( final double max_allowed_gap_ratio,
-                                              final int min_allowed_length,
-                                              final Msa msa ) {
+    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 ) ) );
+            }
+        }
+        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( BasicSequence.copySequence( msa.getSequence( row ) ) );
+            }
+        }
+        if ( seqs.size() < 1 ) {
+            return null;
+        }
+        return BasicMsa.createInstance( seqs );
+    }
+
+    synchronized final public Msa removeGapColumns( final double max_allowed_gap_ratio,
+                                                    final int min_allowed_length,
+                                                    final Msa msa ) {
         init();
         if ( ( max_allowed_gap_ratio < 0 ) || ( max_allowed_gap_ratio > 1 ) ) {
             throw new IllegalArgumentException( "max allowed gap ration is out of range: " + max_allowed_gap_ratio );
@@ -83,7 +109,7 @@ 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;
             }
@@ -153,4 +179,32 @@ 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() );
+    }
 }