inprogress

[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 0f1b961..33254a5 100644 (file)
--- a/forester/java/src/org/forester/application/msa_compactor.java
+++ b/forester/java/src/org/forester/application/msa_compactor.java
@@ -9,47 +9,51 @@ import java.util.List;
 import org.forester.io.parsers.FastaParser;
 import org.forester.io.parsers.GeneralMsaParser;
 import org.forester.msa.Msa;
-import org.forester.msa.MsaCompactor;
+import org.forester.msa.Msa.MSA_FORMAT;
 import org.forester.msa.MsaMethods;
+import org.forester.msa_compactor.MsaCompactor;
 import org.forester.util.CommandLineArguments;
 import org.forester.util.ForesterUtil;
 
 public class msa_compactor {
 
-    final static private String HELP_OPTION_1                 = "help";
-    final static private String HELP_OPTION_2                 = "h";
-    final static private String REMOVE_WORST_OFFENDERS_OPTION = "w";
-    final static private String AV_GAPINESS_OPTION            = "a";
-    final static private String LENGTH_OPTION                 = "l";
-    final static private String REALIGN_OPTION                = "r";
-    final static private String PRG_NAME                      = "msa_compactor";
-    final static private String PRG_DESC                      = "multiple sequnce aligment compactor";
-    final static private String PRG_VERSION                   = "0.90";
-    final static private String PRG_DATE                      = "2012.07.11";
-    final static private String E_MAIL                        = "phylosoft@gmail.com";
-    final static private String WWW                           = "www.phylosoft.org/forester/";
+    final static private String HELP_OPTION_1                          = "help";
+    final static private String HELP_OPTION_2                          = "h";
+    final static private String REMOVE_WORST_OFFENDERS_OPTION          = "w";
+    final static private String AV_GAPINESS_OPTION                     = "a";
+    final static private String STEP_OPTION                            = "s";
+    final static private String LENGTH_OPTION                          = "l";
+    final static private String REALIGN_OPTION                         = "r";
+    final static private String DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION = "nn";
+    final static private String PRG_NAME                               = "msa_compactor";
+    final static private String PRG_DESC                               = "multiple sequnce aligment compactor";
+    final static private String PRG_VERSION                            = "0.01";
+    final static private String PRG_DATE                               = "140314";
+    final static private String E_MAIL                                 = "phylosoft@gmail.com";
+    final static private String WWW                                    = "https://sites.google.com/site/cmzmasek/home/software/forester";
 
     public static void main( final String args[] ) {
         try {
             final CommandLineArguments cla = new CommandLineArguments( args );
-            if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) ) {
+            if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
                 printHelp();
                 System.exit( 0 );
             }
             final File in = cla.getFile( 0 );
+            final File out = cla.getFile( 1 );
             int worst_remove = -1;
             double av = -1;
             int length = -1;
-            final int step = 5;
+            int step = 1;
             boolean realign = false;
-            //            int to = 0;
-            //            int window = 0;
-            //            int step = 0;
+            boolean norm = true;
             final List<String> allowed_options = new ArrayList<String>();
             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
             allowed_options.add( AV_GAPINESS_OPTION );
             allowed_options.add( LENGTH_OPTION );
             allowed_options.add( REALIGN_OPTION );
+            allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
+            allowed_options.add( STEP_OPTION );
             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
             if ( dissallowed_options.length() > 0 ) {
                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
@@ -63,9 +67,15 @@ public class msa_compactor {
             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
             }
+            if ( cla.isOptionSet( STEP_OPTION ) ) {
+                step = cla.getOptionValueAsInt( STEP_OPTION );
+            }
             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
                 realign = true;
             }
+            if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
+                norm = false;
+            }
             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
             //                step = cla.getOptionValueAsInt( STEP_OPTION );
             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
@@ -82,23 +92,21 @@ public class msa_compactor {
             else {
                 msa = GeneralMsaParser.parse( is );
             }
-            System.out.println( msa.toString() );
-            System.out.println( MsaMethods.calcBasicGapinessStatistics( msa ).arithmeticMean() );
             MsaCompactor mc = null;
             if ( worst_remove > 0 ) {
-                mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign );
+                mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign, norm );
             }
             else if ( av > 0 ) {
-                mc = MsaCompactor.reduceGapAverage( msa, av, step, realign );
+                mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50 );
             }
             else if ( length > 0 ) {
                 mc = MsaCompactor.reduceLength( msa, length, step, realign );
             }
-            System.out.println( mc.getMsa().toString() );
-            System.out.println( MsaMethods.calcBasicGapinessStatistics( mc.getMsa() ).arithmeticMean() );
+            System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
             for( final String id : mc.getRemovedSeqIds() ) {
                 System.out.println( id );
             }
+            mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
         }
         catch ( final Exception e ) {
             e.printStackTrace();
@@ -120,11 +128,8 @@ public class msa_compactor {
         System.out.println();
         System.out.println( " options: " );
         System.out.println();
-        //        System.out.println( "   -" + FROM_OPTION + "=<integer>: from (msa column)" );
-        //        System.out.println( "   -" + TO_OPTION + "=<integer>: to (msa column)" );
-        //        System.out.println( "    or" );
-        //        System.out.println( "   -" + WINDOW_OPTION + "=<integer>: window size (msa columns)" );
-        System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>: step size (msa columns)" );
+        System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>  number of sequences to remove" );
+        System.out.println( "   -" + REALIGN_OPTION + " to realign using " );
         System.out.println();
         System.out.println();
         System.out.println();