Merge branch 'master' of gsa2:/humgen/gsa-scr1/chartl/dev/unstable

build.xml

@@ -991,7 +991,7 @@
 
     <!-- Maven install a package consisting of all supporting files. Don't call this target directly. Call one of the specific packaging targets below -->
     <target name="mvninstall" depends="package" description="maven install a package into .m2/repository">
-        <property name="mvn.build.version" value="0.0.1" />
+        <property name="mvn.build.version" value="0.0.2" />
         <!--
         We should use the build version or better yet a git tag version, but tags are currently missing. Alternatively how do we then depend on the LATEST?
         <property name="mvn.build.version" value="${build.version}" />

licensing/private_license.txt

@@ -0,0 +1,7 @@
+PRIVATE PRIVATE PRIVATE
+
+This is a test license for the GATK
+all files will abide to it
+one license to rule them all
+
+PRIVATE PRIVATE PRIVATE

licensing/protected_license.txt

@@ -0,0 +1,7 @@
+PROTECTED PROTECTED PROTECTED
+
+This is a test license for the GATK
+all files will abide to it
+one license to rule them all
+
+PROTECTED PROTECTED PROTECTED

licensing/public_license.txt

@@ -0,0 +1,7 @@
+PUBLIC PUBLIC PUBLIC
+
+This is a test license for the GATK
+all files will abide to it
+one license to rule them all
+
+PUBLIC PUBLIC PUBLIC

protected/java/src/org/broadinstitute/sting/gatk/walkers/haplotypecaller/GenotypingEngine.java

@@ -92,6 +92,7 @@ public class GenotypingEngine {
         cleanUpSymbolicUnassembledEvents( haplotypes );
         if( !in_GGA_mode && samples.size() >= 10 ) { // if not in GGA mode and have at least 10 samples try to create MNP and complex events by looking at LD structure
             mergeConsecutiveEventsBasedOnLD( haplotypes, samples, haplotypeReadMap, startPosKeySet, ref, refLoc );
+            cleanUpSymbolicUnassembledEvents( haplotypes ); // the newly created merged events could be overlapping the unassembled events
         }
         if( in_GGA_mode ) {
             for( final VariantContext compVC : activeAllelesToGenotype ) {
@@ -261,7 +262,6 @@ public class GenotypingEngine {
         return returnMap;
     }
 
-
     protected static void cleanUpSymbolicUnassembledEvents( final List<Haplotype> haplotypes ) {
         final ArrayList<Haplotype> haplotypesToRemove = new ArrayList<Haplotype>();
         for( final Haplotype h : haplotypes ) {
@@ -269,7 +269,7 @@ public class GenotypingEngine {
                 if( vc.isSymbolic() ) {
                     for( final Haplotype h2 : haplotypes ) {
                         for( final VariantContext vc2 : h2.getEventMap().values() ) {
-                            if( vc.getStart() == vc2.getStart() && vc2.isIndel() ) {
+                            if( vc.getStart() == vc2.getStart() && (vc2.isIndel() || vc2.isMNP()) ) { // unfortunately symbolic alleles can't currently be combined with non-point events
                                 haplotypesToRemove.add(h);
                                 break;
                             }

protected/java/src/org/broadinstitute/sting/gatk/walkers/haplotypecaller/HaplotypeCaller.java

@@ -136,6 +136,10 @@ public class HaplotypeCaller extends ActiveRegionWalker<Integer, Integer> implem
     @Argument(fullName="gcpHMM", shortName="gcpHMM", doc="Flat gap continuation penalty for use in the Pair HMM", required = false)
     protected int gcpHMM = 10;
 
+    @Advanced
+    @Argument(fullName="maxNumHaplotypesInPopulation", shortName="maxNumHaplotypesInPopulation", doc="Maximum number of haplotypes to consider for your population. This number will probably need to be increased when calling organisms with high heterozygosity.", required = false)
+    protected int maxNumHaplotypesInPopulation = 13;
+
     @Advanced
     @Argument(fullName="minKmer", shortName="minKmer", doc="Minimum kmer length to use in the assembly graph", required = false)
     protected int minKmer = 11;
@@ -414,7 +418,8 @@ public class HaplotypeCaller extends ActiveRegionWalker<Integer, Integer> implem
         final Map<String, ArrayList<GATKSAMRecord>> perSampleFilteredReadList = splitReadsBySample( filteredReads );
 
         // subset down to only the best haplotypes to be genotyped in all samples ( in GGA mode use all discovered haplotypes )
-        final ArrayList<Haplotype> bestHaplotypes = ( UG_engine.getUAC().GenotypingMode != GenotypeLikelihoodsCalculationModel.GENOTYPING_MODE.GENOTYPE_GIVEN_ALLELES ? likelihoodCalculationEngine.selectBestHaplotypes( haplotypes, stratifiedReadMap ) : haplotypes );
+        final ArrayList<Haplotype> bestHaplotypes = ( UG_engine.getUAC().GenotypingMode != GenotypeLikelihoodsCalculationModel.GENOTYPING_MODE.GENOTYPE_GIVEN_ALLELES ?
+                                                      likelihoodCalculationEngine.selectBestHaplotypes( haplotypes, stratifiedReadMap, maxNumHaplotypesInPopulation ) : haplotypes );
 
         for( final VariantContext call : genotypingEngine.assignGenotypeLikelihoods( UG_engine,
                                                                                      bestHaplotypes,

protected/java/src/org/broadinstitute/sting/gatk/walkers/haplotypecaller/LikelihoodCalculationEngine.java

@@ -273,7 +273,7 @@ public class LikelihoodCalculationEngine {
 
     @Requires({"haplotypes.size() > 0"})
     @Ensures({"result.size() <= haplotypes.size()"})
-    public ArrayList<Haplotype> selectBestHaplotypes( final ArrayList<Haplotype> haplotypes, final Map<String, PerReadAlleleLikelihoodMap> stratifiedReadMap ) {
+    public ArrayList<Haplotype> selectBestHaplotypes( final ArrayList<Haplotype> haplotypes, final Map<String, PerReadAlleleLikelihoodMap> stratifiedReadMap, final int maxNumHaplotypesInPopulation ) {
 
         final int numHaplotypes = haplotypes.size();
         final Set<String> sampleKeySet = stratifiedReadMap.keySet();
@@ -287,7 +287,7 @@ public class LikelihoodCalculationEngine {
         int hap1 = 0;
         int hap2 = 0;
         //double bestElement = Double.NEGATIVE_INFINITY;
-        final int maxChosenHaplotypes = Math.min( 13, sampleKeySet.size() * 2 + 1 );
+        final int maxChosenHaplotypes = Math.min( maxNumHaplotypesInPopulation, sampleKeySet.size() * 2 + 1 );
         while( bestHaplotypesIndexList.size() < maxChosenHaplotypes ) {
             double maxElement = Double.NEGATIVE_INFINITY;
             for( int iii = 0; iii < numHaplotypes; iii++ ) {

public/packages/GATKEngine.xml

@@ -53,11 +53,10 @@
     <file name="public/testdata/exampleFASTA.fasta.fai" />
     <file name="public/testdata/exampleFASTA.dict" />
     <!-- GATK sample code and build scripts -->
-    <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/coverage/DepthOfCoverage.java" />
+    <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/qc/Pileup.java" />
-    <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/Pileup.java" />
-    <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/PrintReads.java" />
     <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/qc/CountLoci.java" />
     <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/qc/CountReads.java" />
     <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/qc/ValidatingPileup.java" />
+    <file name="public/java/src/org/broadinstitute/sting/gatk/walkers/readutils/PrintReads.java" />
   </resources>
 </package>