speed-up for VCF, adding code to the VCF reader to automagically make an index if one doesn't already exist, and a change to the VCF writer unit test

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3305 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/refdata/tracks/builders/TribbleRMDTrackBuilder.java

@@ -61,7 +61,7 @@ public class TribbleRMDTrackBuilder extends PluginManager<FeatureCodec> implemen
 
 
     // the linear index extension
-    private static final String linearIndexExtension = ".idx";
+    public static final String linearIndexExtension = ".idx";
 
     /** Create a new plugin manager. */
     public TribbleRMDTrackBuilder() {
@@ -118,7 +118,7 @@ public class TribbleRMDTrackBuilder extends PluginManager<FeatureCodec> implemen
      * @return a linear index for the specified type
      * @throws IOException if we cannot write the index file
      */
-    private LinearIndex createIndex(File inputFile, FeatureCodec codec) throws IOException {
+    public static LinearIndex createIndex(File inputFile, FeatureCodec codec) throws IOException {
         LinearIndexCreator create = new LinearIndexCreator(inputFile, codec);
         return create.createIndex();
     }
@@ -131,7 +131,7 @@ public class TribbleRMDTrackBuilder extends PluginManager<FeatureCodec> implemen
      * @param inputFile the target file to make an index for
      * @return true if we need to create an index, false otherwise
      */
-    private boolean requireIndex(File inputFile) {
+    public static boolean requireIndex(File inputFile) {
         // can we read the index? if not, create an index
         File indexFile = new File(inputFile.getAbsolutePath() + linearIndexExtension);
         if (!(indexFile.canRead())) return true;

java/src/org/broadinstitute/sting/gatk/refdata/utils/GATKFeature.java

@@ -63,13 +63,15 @@ public abstract class GATKFeature implements Feature {
      */
     public static class TribbleGATKFeature extends GATKFeature {
         private final Feature feature;
+        private GenomeLoc position = null;
         
         public TribbleGATKFeature(Feature f, String name) {
             super(name);
             feature = f;
         }
         public GenomeLoc getLocation() {
-            return GenomeLocParser.createGenomeLoc(feature.getChr(), feature.getStart(), feature.getEnd());
+            if (position == null) position = GenomeLocParser.createGenomeLoc(feature.getChr(), feature.getStart(), feature.getEnd());
+            return position;
         }
 
         /** Return the features reference sequence name, e.g chromosome or contig */

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFReader.java

@@ -21,7 +21,9 @@ import java.util.regex.Pattern;
 import java.util.zip.GZIPInputStream;
 
 import org.broad.tribble.FeatureReader;
+import org.broad.tribble.index.linear.LinearIndex;
 import org.broad.tribble.vcf.*;
+import org.broadinstitute.sting.gatk.refdata.tracks.builders.TribbleRMDTrackBuilder;
 import org.broadinstitute.sting.utils.StingException;
 import org.broadinstitute.sting.utils.Utils;
 
@@ -57,6 +59,14 @@ public class VCFReader implements Iterator<VCFRecord>, Iterable<VCFRecord> {
 
     private void initialize(File vcfFile, VCFCodec.LineTransform transform) {
         VCFCodec codec = new VCFCodec();
+        LinearIndex index = null;
+        if (TribbleRMDTrackBuilder.requireIndex(vcfFile)) {
+            try {
+                index = TribbleRMDTrackBuilder.createIndex(vcfFile, new VCFCodec());
+            } catch (IOException e) {
+                throw new StingException("Unable to make required index for file " + vcfFile + " do you have write permissions to the directory?");
+            }
+        }
         if (transform != null) codec.setTransformer(transform);
         try {
             vcfReader = new FeatureReader(vcfFile,codec);

java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFWriterUnitTest.java

@@ -2,6 +2,7 @@ package org.broadinstitute.sting.utils.genotype.vcf;
 
 import org.broad.tribble.vcf.*;
 import org.broadinstitute.sting.BaseTest;
+import org.broadinstitute.sting.gatk.refdata.tracks.builders.TribbleRMDTrackBuilder;
 import org.broadinstitute.sting.utils.fasta.IndexedFastaSequenceFile;
 import org.broadinstitute.sting.utils.StingException;
 import org.broadinstitute.sting.utils.GenomeLocParser;
@@ -54,6 +55,7 @@ public class VCFWriterUnitTest extends BaseTest {
         }
         Assert.assertEquals(2,counter);
         reader.close();
+        new File(fakeVCFFile + TribbleRMDTrackBuilder.linearIndexExtension).delete();
         fakeVCFFile.delete();
     }