inprogress

diff --git a/forester/java/src/org/forester/msa/BasicMsa.java b/forester/java/src/org/forester/msa/BasicMsa.java
index daeef6b..bda6424 100644 (file)
--- a/forester/java/src/org/forester/msa/BasicMsa.java
+++ b/forester/java/src/org/forester/msa/BasicMsa.java
@@ -192,10 +192,12 @@ public class BasicMsa implements Msa {
         for( int row = 0; row < seqs.size(); ++row ) {
             final Sequence seq = seqs.get( row );
             if ( seq.getLength() != length ) {
-                throw new IllegalArgumentException( "illegal attempt to build msa from sequences of unequal length" );
+                throw new IllegalArgumentException( "illegal attempt to build msa from sequences of unequal length ["
+                        + seq.getIdentifier() + "]" );
             }
             if ( seq.getType() != msa.getType() ) {
-                throw new IllegalArgumentException( "illegal attempt to build msa from sequences of different type" );
+                throw new IllegalArgumentException( "illegal attempt to build msa from sequences of different type ["
+                        + seq.getIdentifier() + "]" );
             }
             if ( ids.contains( seq.getIdentifier() ) ) {
                 throw new IllegalArgumentException( "illegal attempt to create msa with non-unique identifiers ["

diff --git a/forester/java/src/org/forester/msa_compactor/Chart.java b/forester/java/src/org/forester/msa_compactor/Chart.java
index 73f569c..48e959d 100644 (file)
--- a/forester/java/src/org/forester/msa_compactor/Chart.java
+++ b/forester/java/src/org/forester/msa_compactor/Chart.java
@@ -50,10 +50,12 @@ public final class Chart extends JDialog implements ActionListener {
     private ChartPanel          _chart_panel     = null;
     private final JMenuItem     _m_exit          = new JMenuItem();
     private List<MsaProperties> _msa_props;
+    private final int           _initial_number_of_seqs;
 
-    private Chart( final List<MsaProperties> msa_props ) {
+    private Chart( final List<MsaProperties> msa_props, final int initial_number_of_seqs ) {
         super();
         _msa_props = msa_props;
+        _initial_number_of_seqs = initial_number_of_seqs;
         setTitle( "msa compactor" );
         setSize( 500, 400 );
         setResizable( true );
@@ -95,7 +97,7 @@ public final class Chart extends JDialog implements ActionListener {
             }
             final double[][] seqs_length = new double[ _msa_props.size() ][ 2 ];
             for( int i = 0; i < _msa_props.size(); ++i ) {
-                seqs_length[ i ][ 0 ] = _msa_props.get( i ).getNumberOfSequences();
+                seqs_length[ i ][ 0 ] = _initial_number_of_seqs - _msa_props.get( i ).getNumberOfSequences();
                 seqs_length[ i ][ 1 ] = _msa_props.get( i ).getLength();
             }
             model.addData( seqs_length, "Length" );
@@ -103,7 +105,7 @@ public final class Chart extends JDialog implements ActionListener {
             model.setSeriesMarker( "Series " + "Length", false );
             final double[][] seqs_gaps = new double[ _msa_props.size() ][ 2 ];
             for( int i = 0; i < _msa_props.size(); ++i ) {
-                seqs_gaps[ i ][ 0 ] = _msa_props.get( i ).getNumberOfSequences();
+                seqs_gaps[ i ][ 0 ] = _initial_number_of_seqs - _msa_props.get( i ).getNumberOfSequences();
                 seqs_gaps[ i ][ 1 ] = ForesterUtil.roundToInt( _msa_props.get( i ).getGapRatio() * 100 );
             }
             model.addData( seqs_gaps, "Gaps" );
@@ -111,7 +113,7 @@ public final class Chart extends JDialog implements ActionListener {
             model.setSeriesMarker( "Series " + "Gaps", false );
             final double[][] seqs_identity = new double[ _msa_props.size() ][ 2 ];
             for( int i = 0; i < _msa_props.size(); ++i ) {
-                seqs_identity[ i ][ 0 ] = _msa_props.get( i ).getNumberOfSequences();
+                seqs_identity[ i ][ 0 ] = _initial_number_of_seqs - _msa_props.get( i ).getNumberOfSequences();
                 seqs_identity[ i ][ 1 ] = ForesterUtil.roundToInt( _msa_props.get( i ).getAverageIdentityRatio() * 100 );
             }
             model.addData( seqs_identity, "Id" );
@@ -131,14 +133,14 @@ public final class Chart extends JDialog implements ActionListener {
         return _chart_panel;
     }
 
-    public static void display( final List<MsaProperties> msa_props ) {
+    public static void display( final List<MsaProperties> msa_props, final int initial_number_of_seqs ) {
         try {
             UIManager.setLookAndFeel( UIManager.getSystemLookAndFeelClassName() );
         }
         catch ( final Exception e ) {
             e.printStackTrace();
         }
-        final Chart chart = new Chart( msa_props );
+        final Chart chart = new Chart( msa_props, initial_number_of_seqs );
         chart.setVisible( true );
     }
 
@@ -149,7 +151,7 @@ public final class Chart extends JDialog implements ActionListener {
         catch ( final Exception e ) {
             e.printStackTrace();
         }
-        final Chart temp = new Chart( null );
+        final Chart temp = new Chart( null, 0 );
         temp.setVisible( true );
     }
 }

diff --git a/forester/java/src/org/forester/msa_compactor/MsaCompactor.java b/forester/java/src/org/forester/msa_compactor/MsaCompactor.java
index 210e9e9..0ba6b88 100644 (file)
--- a/forester/java/src/org/forester/msa_compactor/MsaCompactor.java
+++ b/forester/java/src/org/forester/msa_compactor/MsaCompactor.java
@@ -172,22 +172,32 @@ public class MsaCompactor {
             printTableHeader();
         }
         int i = 0;
-        final int x = ForesterUtil.roundToInt( _msa.getNumberOfSequences() / 20.0 );
+        final int s = _msa.getNumberOfSequences();
+        final int x = ForesterUtil.roundToInt( s / 20.0 );
         while ( _msa.getNumberOfSequences() > x ) {
             final String id = to_remove_ids.get( i );
             _msa = MsaMethods.removeSequence( _msa, id );
-            removeGapColumns();
-            msa_props.add( new MsaProperties( _msa ) );
-            if ( verbose ) {
-                printMsaStats( id );
-            }
-            if ( ( step > 0 ) && ( ( ( i + 1 ) % step ) == 0 ) ) {
-                if ( realign ) {
+            if ( ( s < 500 ) || ( ( step > 0 ) && ( ( ( i + 1 ) % step ) == 0 ) ) ) {
+                removeGapColumns();
+                if ( realign && ( ( step > 0 ) && ( ( ( i + 1 ) % step ) == 0 ) ) ) {
                     realignWithMafft();
+                    msa_props.add( new MsaProperties( _msa ) );
+                    if ( verbose ) {
+                        printMsaStats( id );
+                    }
+                    if ( verbose ) {
+                        System.out.print( "(realigned)" );
+                    }
+                }
+                else {
+                    msa_props.add( new MsaProperties( _msa ) );
+                    if ( verbose ) {
+                        printMsaStats( id );
+                    }
+                }
+                if ( verbose ) {
+                    System.out.println();
                 }
-            }
-            if ( verbose ) {
-                System.out.println();
             }
             ++i;
         }
@@ -256,7 +266,7 @@ public class MsaCompactor {
         }
         final String s = writeOutfile();
         if ( verbose ) {
-            System.out.print( "-> " + s + ( realign ? " (realigned)" : "" ) );
+            System.out.print( "-> " + s + ( realign ? "\t(realigned)" : "" ) );
         }
     }
 
@@ -393,12 +403,13 @@ public class MsaCompactor {
                                             final boolean realign,
                                             final boolean norm,
                                             final String path_to_mafft ) throws IOException, InterruptedException {
+        final int initial_number_of_seqs = msa.getNumberOfSequences();
         final MsaCompactor mc = new MsaCompactor( msa );
         if ( realign ) {
             mc.setPathToMafft( path_to_mafft );
         }
         final List<MsaProperties> msa_props = mc.chart( step, realign, norm, true );
-        Chart.display( msa_props );
+        Chart.display( msa_props, initial_number_of_seqs );
         return mc;
     }