in progress

[jalview.git] / forester / java / src / org / forester / application / msa_compactor.java

diff --git a/forester/java/src/org/forester/application/msa_compactor.java b/forester/java/src/org/forester/application/msa_compactor.java
index 15c2d4c..e555593 100644 (file)
--- a/forester/java/src/org/forester/application/msa_compactor.java
+++ b/forester/java/src/org/forester/application/msa_compactor.java
@@ -105,7 +105,7 @@ public class msa_compactor {
             int length = -1;
             int step = 1;
             boolean realign = false;
-            boolean norm = true;
+            boolean normalize_for_effective_seq_length = true;
             String path_to_mafft = null;
             int step_for_diagnostics = 1;
             int min_length = -1;
@@ -223,7 +223,7 @@ public class msa_compactor {
                 path_to_mafft = cla.getOptionValueAsCleanString( PATH_TO_MAFFT_OPTION );
             }
             if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
-                norm = false;
+                normalize_for_effective_seq_length = false;
             }
             if ( cla.isOptionSet( STEP_FOR_DIAGNOSTICS_OPTION ) ) {
                 step_for_diagnostics = cla.getOptionValueAsInt( STEP_FOR_DIAGNOSTICS_OPTION );
@@ -346,8 +346,11 @@ public class msa_compactor {
                 }
                 System.out.println( "Step for diagnostics reports         : " + step_for_diagnostics );
                 System.out.println( "Calculate normalized Shannon Entropy : " + report_entropy );
-                if ( !norm ) {
-                    System.out.println( "Normalize                            : " + norm );
+                if ( normalize_for_effective_seq_length ) {
+                    System.out.println( "Normalize                            : with individual, effective sequence lenghts"  );
+                }
+                else {
+                    System.out.println( "Normalize                            : with MSA length" );
                 }
                 System.out.println( "Realign with MAFFT                   : " + realign );
                 if ( realign ) {
@@ -372,7 +375,7 @@ public class msa_compactor {
                 if ( removed_seqs_out_base != null ) {
                     mc.setRemovedSeqsOutBase( removed_seqs_out_base );
                 }
-                mc.setNorm( norm );
+                mc.setNorm( normalize_for_effective_seq_length );
                 mc.setRealign( realign );
                 if ( realign ) {
                     mc.setPathToMafft( path_to_mafft );
@@ -391,9 +394,12 @@ public class msa_compactor {
                     msa_props = mc.removeViaLength( length );
                 }
                 else {
-                    msa_props = mc.chart( step, realign, norm );
+                    msa_props = mc.chart( step, realign, normalize_for_effective_seq_length );
                 }
                 Chart.display( msa_props, initial_number_of_seqs, report_entropy, in.getName() );
+                System.out.println();
+                System.out.println( "Final MSA properties" );
+                printMsaInfo( msa,  MsaMethods.calculateEffectiveLengthStatistics( msa ));
             }
         }
         catch ( final IllegalArgumentException iae ) {
@@ -418,20 +424,24 @@ public class msa_compactor {
                                               WWW,
                                               ForesterUtil.getForesterLibraryInformation() );
         System.out.println( "Input MSA                            : " + in );
-        System.out.println( "  MSA length                         : " + msa.getLength() );
-        System.out.println( "  Number of sequences                : " + msa.getNumberOfSequences() );
-        System.out.println( "  Median sequence length             : " + NF_1.format( initial_msa_stats.median() ) );
-        System.out.println( "  Mean sequence length               : "
-                + NF_1.format( initial_msa_stats.arithmeticMean() ) );
-        System.out.println( "  Max sequence length                : " + ( ( int ) initial_msa_stats.getMax() ) );
-        System.out.println( "  Min sequence length                : " + ( ( int ) initial_msa_stats.getMin() ) );
-        System.out.println( "  Gap ratio                          : "
+        printMsaInfo( msa, initial_msa_stats );
+    }
+
+    private static void printMsaInfo( DeleteableMsa msa, final DescriptiveStatistics msa_stats ) {
+        System.out.println( "MSA length                           : " + msa.getLength() );
+        System.out.println( "Number of sequences                  : " + msa.getNumberOfSequences() );
+        System.out.println( "Median sequence length               : " + NF_1.format( msa_stats.median() ) );
+        System.out.println( "Mean sequence length                 : "
+                + NF_1.format( msa_stats.arithmeticMean() ) );
+        System.out.println( "Max sequence length                  : " + ( ( int ) msa_stats.getMax() ) );
+        System.out.println( "Min sequence length                  : " + ( ( int ) msa_stats.getMin() ) );
+        System.out.println( "Gap ratio                            : "
                 + NF_4.format( MsaMethods.calcGapRatio( msa ) ) );
-        System.out.println( "  Mean gap count per 100 residues    : "
-                + NF_1.format( MsaMethods.calcNumberOfGapsPer100Stats( msa ).arithmeticMean() ) );
-        System.out.println( "  Normalized Shannon Entropy (entn7) : "
+        System.out.println( "Mean gap count per sequence          : "
+                + NF_1.format( MsaMethods.calcNumberOfGapsStats( msa ).arithmeticMean() ) );
+        System.out.println( "Normalized Shannon Entropy (entn7)   : "
                 + NF_4.format( MsaMethods.calcNormalizedShannonsEntropy( 7, msa ) ) );
-        System.out.println( "  Normalized Shannon Entropy (entn21): "
+        System.out.println( "Normalized Shannon Entropy (entn21)  : "
                 + NF_4.format( MsaMethods.calcNormalizedShannonsEntropy( 21, msa ) ) );
     }
 
@@ -493,6 +503,9 @@ public class msa_compactor {
                             + "=<decimal>  maximal allowed gap ratio per column (for deleting of columms) (0.0-1.0)" );
         System.out.println( "   -" + PERFORM_PHYLOGENETIC_INFERENCE
                             + "             to calculate a simple phylogenetic tree (Kimura distances, NJ)" );
+        System.out.println( "   -" + DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION
+                            + "            to normalize gap-contributions with MSA length, instead of individual effective sequence lenghts" );
+       
         System.out.println();
         System.out.println();
         System.out.println();