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 765648f..1cd94d6 100644 (file)
--- a/forester/java/src/org/forester/application/msa_compactor.java
+++ b/forester/java/src/org/forester/application/msa_compactor.java
@@ -9,7 +9,6 @@ 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.Msa.MSA_FORMAT;
 import org.forester.msa.MsaInferrer;
 import org.forester.msa_compactor.MsaCompactor;
 import org.forester.util.CommandLineArguments;
@@ -27,9 +26,9 @@ public class msa_compactor {
     final static private String PATH_TO_MAFFT_OPTION                   = "mafft";
     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_DESC                               = "multiple sequence aligment compactor";
     final static private String PRG_VERSION                            = "0.01";
-    final static private String PRG_DATE                               = "140314";
+    final static private String PRG_DATE                               = "140316";
     final static private String E_MAIL                                 = "phylosoft@gmail.com";
     final static private String WWW                                    = "https://sites.google.com/site/cmzmasek/home/software/forester";
 
@@ -43,9 +42,9 @@ public class msa_compactor {
             final File in = cla.getFile( 0 );
             final File out = cla.getFile( 1 );
             int worst_remove = -1;
-            double av = -1;
+            double av_gap = -1;
             int length = -1;
-            int step = 1;
+            int step = -1;
             boolean realign = false;
             boolean norm = true;
             String path_to_mafft = null;
@@ -61,17 +60,47 @@ public class msa_compactor {
             if ( dissallowed_options.length() > 0 ) {
                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
             }
+            Msa msa = null;
+            final FileInputStream is = new FileInputStream( in );
+            if ( FastaParser.isLikelyFasta( in ) ) {
+                msa = FastaParser.parseMsa( is );
+            }
+            else {
+                msa = GeneralMsaParser.parse( is );
+            }
             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
+                if ( ( worst_remove < 1 ) || ( worst_remove >= msa.getNumberOfSequences() - 1 ) ) {
+                    ForesterUtil.fatalError( PRG_NAME, "number of worst offender sequences to remove is out of range: "
+                            + worst_remove );
+                }
             }
             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
-                av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
+                if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
+                    printHelp();
+                    System.exit( 0 );
+                }
+                av_gap = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
+                if ( ( av_gap < 0 ) || ( av_gap >= 1 ) ) {
+                    ForesterUtil.fatalError( PRG_NAME, "target gap-ratio is out of range: " + av_gap );
+                }
             }
             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
+                if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) || cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
+                    printHelp();
+                    System.exit( 0 );
+                }
                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
+                if ( ( length < 2 ) || ( length >= msa.getLength() ) ) {
+                    ForesterUtil.fatalError( PRG_NAME, "target length is out of range: " + length );
+                }
             }
             if ( cla.isOptionSet( STEP_OPTION ) ) {
                 step = cla.getOptionValueAsInt( STEP_OPTION );
+                if ( ( step < 1 )
+                        || ( ( step > msa.getNumberOfSequences() ) || ( ( worst_remove > 0 ) && ( step > worst_remove ) ) ) ) {
+                    ForesterUtil.fatalError( PRG_NAME, "value for step is out of range: " + step );
+                }
             }
             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
                 realign = true;
@@ -85,43 +114,22 @@ public class msa_compactor {
             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 );
-            //            }
-            //            else {
-            //                printHelp();
-            //                System.exit( 0 );
-            //            }
             if ( realign ) {
                 if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
                     path_to_mafft = MsaCompactor.guessPathToMafft();
                 }
                 checkPathToMafft( path_to_mafft );
             }
-            Msa msa = null;
-            final FileInputStream is = new FileInputStream( in );
-            if ( FastaParser.isLikelyFasta( in ) ) {
-                msa = FastaParser.parseMsa( is );
-            }
-            else {
-                msa = GeneralMsaParser.parse( is );
-            }
-            MsaCompactor mc = null;
             if ( worst_remove > 0 ) {
-                mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign, norm, path_to_mafft );
+                MsaCompactor.removeWorstOffenders( msa, worst_remove, step, realign, norm, path_to_mafft, out );
             }
-            else if ( av > 0 ) {
-                mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50, path_to_mafft );
+            else if ( av_gap > 0 ) {
+                MsaCompactor.reduceGapAverage( msa, av_gap, step, realign, norm, path_to_mafft, out );
             }
             else if ( length > 0 ) {
-                mc = MsaCompactor.reduceLength( msa, length, step, realign, path_to_mafft );
+                // TODO if < shortest seq -> error
+                MsaCompactor.reduceLength( msa, length, step, realign, norm, path_to_mafft, out );
             }
-            //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();
@@ -168,6 +176,9 @@ public class msa_compactor {
         System.out.println();
         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION
                 + "=<integer>  number of worst offender sequences to remove" );
+        System.out.println( "   -" + LENGTH_OPTION + "=<integer>  target MSA length" );
+        System.out.println( "   -" + AV_GAPINESS_OPTION + "=<decimal>  target gap-ratio (0.0-1.0)" );
+        System.out.println( "   -" + STEP_OPTION + "=<integer>  step (for output and re-aligning)" );
         System.out.println( "   -" + REALIGN_OPTION + "            to realign using MAFFT" + mafft_comment );
         System.out.println();
         System.out.println();