Change the interface for RMDTrackBuilder, therefore always mandating the specification

of a sequence dictionary and related info.  This will hopefully eliminate the cases in
which the refseq track depends a sequence dictionary / contig parser that hasn't been
specified.


git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4700 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/AbstractGenomeAnalysisEngine.java

@@ -326,11 +326,8 @@ public abstract class AbstractGenomeAnalysisEngine {
         
         sampleDataSource = new SampleDataSource(getSAMFileHeader(), argCollection.sampleFiles);
 
-        RMDTrackBuilder manager = new RMDTrackBuilder();
-
         // set the sequence dictionary of all of Tribble tracks to the sequence dictionary of our reference
-        manager.setSequenceDictionary(referenceDataSource.getReference().getSequenceDictionary(),genomeLocParser,argCollection.unsafe);
+        RMDTrackBuilder manager = new RMDTrackBuilder(referenceDataSource.getReference().getSequenceDictionary(),genomeLocParser,argCollection.unsafe);
-
         List<RMDTrack> tracks = manager.getReferenceMetaDataSources(this,argCollection);
         validateSuppliedReferenceOrderedData(tracks);
 
@@ -591,7 +588,11 @@ public abstract class AbstractGenomeAnalysisEngine {
     private List<ReferenceOrderedDataSource> getReferenceOrderedDataSources(List<RMDTrack> rods) {
         List<ReferenceOrderedDataSource> dataSources = new ArrayList<ReferenceOrderedDataSource>();
         for (RMDTrack rod : rods)
-            dataSources.add(new ReferenceOrderedDataSource(referenceDataSource.getReference().getSequenceDictionary(),genomeLocParser,rod,flashbackData()));
+            dataSources.add(new ReferenceOrderedDataSource(referenceDataSource.getReference().getSequenceDictionary(),
+                                                           genomeLocParser,
+                                                           argCollection.unsafe,
+                                                           rod,
+                                                           flashbackData()));
         return dataSources;
     }
 

java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/ReferenceOrderedDataSource.java

@@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.datasources.simpleDataSources;
 
 import net.sf.samtools.SAMSequenceDictionary;
 import org.broad.tribble.FeatureSource;
+import org.broadinstitute.sting.gatk.arguments.ValidationExclusion;
 import org.broadinstitute.sting.gatk.datasources.shards.Shard;
 import org.broadinstitute.sting.gatk.refdata.SeekableRODIterator;
 import org.broadinstitute.sting.gatk.refdata.tracks.RMDTrack;
@@ -49,10 +50,16 @@ public class ReferenceOrderedDataSource implements SimpleDataSource {
      * Create a new reference-ordered data source.
      * @param rod the reference ordered data
      */
-    public ReferenceOrderedDataSource(SAMSequenceDictionary sequenceDictionary,GenomeLocParser genomeLocParser, RMDTrack rod, boolean flashbackData ) {
+    public ReferenceOrderedDataSource(SAMSequenceDictionary sequenceDictionary,
+                                      GenomeLocParser genomeLocParser,
+                                      ValidationExclusion.TYPE validationExclusionType,
+                                      RMDTrack rod, boolean flashbackData ) {
         this.rod = rod;
         if (rod.supportsQuery())
-            iteratorPool = new ReferenceOrderedQueryDataPool(sequenceDictionary,genomeLocParser,new RMDTrackBuilder(),rod);
+            iteratorPool = new ReferenceOrderedQueryDataPool(sequenceDictionary,
+                                                             genomeLocParser,
+                                                             new RMDTrackBuilder(sequenceDictionary,genomeLocParser,validationExclusionType),
+                                                             rod);
         else
             iteratorPool = new ReferenceOrderedDataPool(sequenceDictionary,genomeLocParser,rod, flashbackData );
     }

java/src/org/broadinstitute/sting/gatk/refdata/indexer/RMDIndexer.java

@@ -11,6 +11,7 @@ import org.broadinstitute.sting.commandline.Argument;
 import org.broadinstitute.sting.commandline.CommandLineProgram;
 import org.broadinstitute.sting.commandline.Input;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
+import org.broadinstitute.sting.gatk.arguments.ValidationExclusion;
 import org.broadinstitute.sting.gatk.refdata.ReferenceDependentFeatureCodec;
 import org.broadinstitute.sting.gatk.refdata.tracks.builders.RMDTrackBuilder;
 import org.broadinstitute.sting.utils.GenomeLocParser;
@@ -79,7 +80,7 @@ public class RMDIndexer extends CommandLineProgram {
         genomeLocParser = new GenomeLocParser(ref);
 
         // get a track builder
-        RMDTrackBuilder builder = new RMDTrackBuilder();
+        RMDTrackBuilder builder = new RMDTrackBuilder(ref.getSequenceDictionary(),genomeLocParser, ValidationExclusion.TYPE.ALL);
 
         // find the types available to the track builders
         Map<String,Class> typeMapping = builder.getAvailableTrackNamesAndTypes();
@@ -100,13 +101,8 @@ public class RMDIndexer extends CommandLineProgram {
 
         Index index = IndexFactory.createIndex(inputFileSource, codec, approach);
 
-        if (referenceFile != null) {
+        // add writing of the sequence dictionary, if supplied
-            // create a reference file reader
+        builder.setIndexSequenceDictionary(inputFileSource, index, ref.getSequenceDictionary(), indexFile, false);
-            IndexedFastaSequenceFile seq = new IndexedFastaSequenceFile(referenceFile);
-
-            // add writing of the sequence dictionary, if supplied
-            builder.setIndexSequenceDictionary(inputFileSource, index, seq.getSequenceDictionary(), indexFile, false);
-        }
 
         // create the output stream, and write the index
         LittleEndianOutputStream stream = new LittleEndianOutputStream(new FileOutputStream(indexFile));

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

@@ -92,11 +92,6 @@ public class RMDTrackBuilder extends PluginManager<FeatureCodec> {
      */
     private ValidationExclusion.TYPE validationExclusionType;
 
-    /** Create a new plugin manager. */
-    public RMDTrackBuilder() {
-        super(FeatureCodec.class, "Codecs", "Codec");
-    }
-
     /**
      * Create a new RMDTrackBuilder, with dictionary and genomeLocParser predefined.
      * @param dict Sequence dictionary to use.
@@ -105,20 +100,10 @@ public class RMDTrackBuilder extends PluginManager<FeatureCodec> {
      */
     public RMDTrackBuilder(SAMSequenceDictionary dict,GenomeLocParser genomeLocParser, ValidationExclusion.TYPE validationExclusionType) {
         super(FeatureCodec.class, "Codecs", "Codec");
-        setSequenceDictionary(dict,genomeLocParser,validationExclusionType);
-    }
-
-    /**
-     * Establish location-aware parsing and services for relevant reference metadata.
-     * @param dict Sequence dictionary to use.
-     * @param genomeLocParser Location parser to use.
-     * @param validationExclusionType Types of validations to exclude, for sequence dictionary verification.
-     */
-    public void setSequenceDictionary(SAMSequenceDictionary dict,GenomeLocParser genomeLocParser,ValidationExclusion.TYPE validationExclusionType) {
         this.dict = dict;
         this.genomeLocParser = genomeLocParser;
         this.validationExclusionType = validationExclusionType;
-    }    
+    }
 
     /** @return a list of all available track types we currently have access to create */
     public Map<String, Class> getAvailableTrackNamesAndTypes() {

java/src/org/broadinstitute/sting/gatk/walkers/indels/IndelGenotyperV2Walker.java

@@ -248,7 +248,9 @@ public class IndelGenotyperV2Walker extends ReadWalker<Integer,Integer> {
 		if ( RefseqFileName != null ) {
             logger.info("Using RefSeq annotations from "+RefseqFileName);
 
-			RMDTrackBuilder builder = new RMDTrackBuilder();
+			RMDTrackBuilder builder = new RMDTrackBuilder(getToolkit().getReferenceDataSource().getReference().getSequenceDictionary(),
+                                                          getToolkit().getGenomeLocParser(),
+                                                          getToolkit().getArguments().unsafe);
             FeatureSource refseq = builder.createFeatureReader(RefSeqCodec.class,new File(RefseqFileName)).first;
 
             try {

java/src/org/broadinstitute/sting/oneoffprojects/walkers/DbSNPWindowCounter.java

@@ -47,7 +47,9 @@ public class DbSNPWindowCounter extends LocusWalker<Integer, Long> {
 
 
     public void initialize() {
-        RMDTrackBuilder builder = new RMDTrackBuilder();
+        RMDTrackBuilder builder = new RMDTrackBuilder(getToolkit().getReferenceDataSource().getReference().getSequenceDictionary(),
+                                                      getToolkit().getGenomeLocParser(),
+                                                      getToolkit().getArguments().unsafe);
         reader = builder.createFeatureReader(DbSNPCodec.class,myDbSNPFile).first;
     }
 

java/src/org/broadinstitute/sting/oneoffprojects/walkers/IndelAnnotator.java

@@ -35,7 +35,9 @@ public class IndelAnnotator extends RodWalker<Integer,Long> {
 
     public void initialize() {
         if ( RefseqFileName != null ) {
-            RMDTrackBuilder builder = new RMDTrackBuilder();
+            RMDTrackBuilder builder = new RMDTrackBuilder(getToolkit().getReferenceDataSource().getReference().getSequenceDictionary(),
+                                                          getToolkit().getGenomeLocParser(),
+                                                          getToolkit().getArguments().unsafe);
             FeatureSource refseq = builder.createFeatureReader(RefSeqCodec.class,new File(RefseqFileName)).first;
 
             try {

java/src/org/broadinstitute/sting/queue/extensions/gatk/GATKExtensionsGenerator.java

@@ -32,6 +32,7 @@ import org.broadinstitute.sting.commandline.*;
 import org.broadinstitute.sting.gatk.CommandLineGATK;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
 import org.broadinstitute.sting.gatk.WalkerManager;
+import org.broadinstitute.sting.gatk.arguments.ValidationExclusion;
 import org.broadinstitute.sting.gatk.filters.FilterManager;
 import org.broadinstitute.sting.gatk.io.stubs.VCFWriterArgumentTypeDescriptor;
 import org.broadinstitute.sting.gatk.io.stubs.OutputStreamArgumentTypeDescriptor;
@@ -67,7 +68,10 @@ public class GATKExtensionsGenerator extends CommandLineProgram {
     GenomeAnalysisEngine GATKEngine = new GenomeAnalysisEngine();
     WalkerManager walkerManager = new WalkerManager();
     FilterManager filterManager = new FilterManager();
-    RMDTrackBuilder trackBuilder = new RMDTrackBuilder();
+    // HACK: We're currently relying on the fact that RMDTrackBuilder is used only from RMD type lookups, not
+    //       RMD track location.  Therefore, no sequence dictionary is required.  In the future, we should separate
+    //       RMD track lookups from track creation.
+    RMDTrackBuilder trackBuilder = new RMDTrackBuilder(null,null,ValidationExclusion.TYPE.ALL);
 
     /**
      * Required main method implementation.

java/test/org/broadinstitute/sting/gatk/datasources/providers/ReferenceOrderedViewUnitTest.java

@@ -55,8 +55,7 @@ public class ReferenceOrderedViewUnitTest extends BaseTest {
         // sequence
         seq = new IndexedFastaSequenceFile(new File(hg18Reference));
         genomeLocParser = new GenomeLocParser(seq);
-        builder = new RMDTrackBuilder();
+        builder = new RMDTrackBuilder(seq.getSequenceDictionary(),genomeLocParser,null);
-        builder.setSequenceDictionary(seq.getSequenceDictionary(),genomeLocParser,null);
     }
 
     /**
@@ -79,7 +78,7 @@ public class ReferenceOrderedViewUnitTest extends BaseTest {
     public void testSingleBinding() {
         File file = new File(testDir + "TabularDataTest.dat");
         RMDTrack track = builder.createInstanceOfTrack(TableCodec.class,"tableTest",file);
-        ReferenceOrderedDataSource dataSource = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,track,false);
+        ReferenceOrderedDataSource dataSource = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,null,track,false);
 
         Shard shard = new MockLocusShard(genomeLocParser,Collections.singletonList(genomeLocParser.createGenomeLoc("chrM",1,30)));
 
@@ -103,9 +102,9 @@ public class ReferenceOrderedViewUnitTest extends BaseTest {
 
 
         RMDTrack track = builder.createInstanceOfTrack(TableCodec.class,"tableTest1",file);
-        ReferenceOrderedDataSource dataSource1 = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,track,false);
+        ReferenceOrderedDataSource dataSource1 = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,null,track,false);
         RMDTrack track2 = builder.createInstanceOfTrack(TableCodec.class,"tableTest2",file);
-        ReferenceOrderedDataSource dataSource2 = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,track2,false);
+        ReferenceOrderedDataSource dataSource2 = new ReferenceOrderedDataSource(seq.getSequenceDictionary(),genomeLocParser,null,track2,false);
 
 
         Shard shard = new MockLocusShard(genomeLocParser,Collections.singletonList(genomeLocParser.createGenomeLoc("chrM",1,30)));

java/test/org/broadinstitute/sting/gatk/datasources/simpleDataSources/ReferenceOrderedDataPoolUnitTest.java

@@ -60,8 +60,7 @@ public class ReferenceOrderedDataPoolUnitTest extends BaseTest {
     @BeforeMethod
     public void setUp() {
         File file = new File(testDir + "TabularDataTest.dat");
-        RMDTrackBuilder builder = new RMDTrackBuilder();
+        RMDTrackBuilder builder = new RMDTrackBuilder(seq.getSequenceDictionary(),genomeLocParser,null);
-        builder.setSequenceDictionary(seq.getSequenceDictionary(),genomeLocParser,null);        
         rod = builder.createInstanceOfTrack(TableCodec.class, "tableTest", file);
     }
 

java/test/org/broadinstitute/sting/gatk/refdata/tracks/builders/RMDTrackBuilderUnitTest.java

@@ -60,7 +60,7 @@ public class RMDTrackBuilderUnitTest extends BaseTest {
     public void setup() {
         seq = new IndexedFastaSequenceFile(new File(b36KGReference));
         genomeLocParser = new GenomeLocParser(seq);
-        builder = new RMDTrackBuilder();
+        builder = new RMDTrackBuilder(seq.getSequenceDictionary(),genomeLocParser,null);
     }
 
     @Test
@@ -153,7 +153,6 @@ public class RMDTrackBuilderUnitTest extends BaseTest {
         File vcfFile = createCorrectDateIndexFile(new File(validationDataLocation + "/ROD_validation/newerTribbleTrack.vcf"));
         Long indexTimeStamp = Tribble.indexFile(vcfFile).lastModified();
         try {
-            builder.setSequenceDictionary(seq.getSequenceDictionary(),genomeLocParser,null);
             Index idx = builder.loadIndex(vcfFile, new VCFCodec());
             // catch any exception; this call should pass correctly
             SAMSequenceDictionary dict =  RMDTrackBuilder.getSequenceDictionaryFromProperties(idx);