-Misc improvements to VCF code

-Small fix to callset concordance


git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2497 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/walkers/concordance/NWayVenn.java

@@ -25,7 +25,8 @@ public class NWayVenn implements ConcordanceType {
 
         TreeSet<String> concordantSamples = new TreeSet<String>();
         for ( Entry<String, Genotype> entry : samplesToRecords.entrySet() ) {
-            concordantSamples.add(entry.getKey());
+            if ( !entry.getValue().isNoCall() )
+                concordantSamples.add(entry.getKey());
         }
 
         StringBuffer tag = new StringBuffer();

java/src/org/broadinstitute/sting/gatk/walkers/concordance/SNPGenotypeConcordance.java

@@ -34,7 +34,11 @@ public class SNPGenotypeConcordance implements ConcordanceType {
         char refBase = ref.getBase();
 
         Genotype call1 = samplesToRecords.get(sample1);
+        if ( call1 != null && call1.isNoCall() )
+            call1 = null;
         Genotype call2 = samplesToRecords.get(sample2);
+        if ( call2 != null && call2.isNoCall() )
+            call2 = null;
 
         if ( call1 == null || call2 == null ) {
             if ( call1 != null && call1.isPointGenotype() && call1.isVariant(refBase) ) {

java/src/org/broadinstitute/sting/gatk/walkers/concordance/SimpleVenn.java

@@ -29,7 +29,11 @@ public class SimpleVenn implements ConcordanceType {
     public String computeConcordance(Map<String, Genotype> samplesToRecords, ReferenceContext ref) {
 
         Genotype call1 = samplesToRecords.get(sample1);
+        if ( call1 != null && call1.isNoCall() )
+            call1 = null;
         Genotype call2 = samplesToRecords.get(sample2);
+        if ( call2 != null && call2.isNoCall() )
+            call2 = null;
 
         if ( call1 == null && call2 == null )
             return null;

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

@@ -247,19 +247,23 @@ public class VCFGenotypeRecord implements Genotype, SampleBacked {
     public String toStringEncoding(List<VCFGenotypeEncoding> altAlleles, String[] genotypeFormatStrings) {
         StringBuilder builder = new StringBuilder();
         builder.append(toGenotypeString(altAlleles));
-        for ( String field : genotypeFormatStrings ) {
+
-            String value = mFields.get(field);
+        if ( !isEmptyGenotype() ) {
-            if ( value == null && field.equals(OLD_DEPTH_KEY) )
+            for ( String field : genotypeFormatStrings ) {
+                String value = mFields.get(field);
+                if ( value == null && field.equals(OLD_DEPTH_KEY) )
                     value = mFields.get(DEPTH_KEY);
-            if ( value == null )
+                if ( value == null )
                     continue;
 
-            builder.append(VCFRecord.GENOTYPE_FIELD_SEPERATOR);
+                builder.append(VCFRecord.GENOTYPE_FIELD_SEPERATOR);
-            if (value.equals(""))
+                if (value.equals(""))
-                builder.append(getMissingFieldValue(field));
+                    builder.append(getMissingFieldValue(field));
-            else
+                else
-                builder.append(value);
+                    builder.append(value);
+            }
         }
+
         return builder.toString();
     }
 

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

@@ -139,6 +139,8 @@ public class VCFUtils {
         int SLODsSeen = 0;
         double totalFreq = 0.0;
         int freqsSeen = 0;
+        String id = null;
+        String filter = null;
 
         for ( RodVCF rod : rods ) {
             List<VCFGenotypeRecord> myGenotypes = rod.getVCFGenotypeRecords();
@@ -158,7 +160,7 @@ public class VCFUtils {
             if ( confidence > maxConfidence )
                 maxConfidence = confidence;
 
-            if ( rod.hasNonRefAlleleFrequency() ) {
+            if ( !rod.isReference() && rod.hasNonRefAlleleFrequency() ) {
                 totalFreq += rod.getNonRefAlleleFrequency();
                 freqsSeen++;
             }
@@ -167,6 +169,12 @@ public class VCFUtils {
                 totalSLOD += rod.getStrandBias();
                 SLODsSeen++;
             }
+
+            if ( rod.getID() != null )
+                id = rod.getID();
+
+            if ( rod.hasFilteringCodes() )
+                filter = rod.getFilterString();
         }
 
         Map<String, String> infoFields = new HashMap<String, String>();
@@ -178,16 +186,14 @@ public class VCFUtils {
             infoFields.put(VCFRecord.STRAND_BIAS_KEY, String.format("%.2f", (totalSLOD/(double)SLODsSeen)));
         if ( freqsSeen > 0 )
             infoFields.put(VCFRecord.ALLELE_FREQUENCY_KEY, String.format("%.2f", (totalFreq/(double)freqsSeen)));
-
+                
-        // TODO -- "." and "0" are wrong -- need to use values from the records
-
         return new VCFRecord(params.getReferenceBase(),
                 params.getContig(),
                 params.getPosition(),
-                ".",
+                (id != null ? id : "."),
                 params.getAlternateBases(),
                 maxConfidence,
-                "0",
+                (filter != null ? filter : "."),
                 infoFields,
                 params.getFormatString(),
                 params.getGenotypesRecords());

java/test/org/broadinstitute/sting/gatk/walkers/concordance/CallsetConcordanceIntegrationTest.java

@@ -30,7 +30,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
     public void testNWayVenn() {
         WalkerTestSpec spec = new WalkerTestSpec(
                 baseTestString() + " -B set1,VCF," + validationDataLocation + "NA12878.example1.vcf -B set2,VCF," + validationDataLocation + "NA12878.example2.vcf -B set3,VCF," + validationDataLocation + "CEU.sample.vcf -CT NWayVenn", 1,
-                Arrays.asList("0527ea8ec7de3a144bd0a56db80d62ba"));
+                Arrays.asList("86d2342fabc8c0916a6d42a29f750ea2"));
         executeTest("testNWayVenn", spec);
     }
 
@@ -41,4 +41,12 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
                 Arrays.asList("6fbe00cb68d2cdc59dfcb79024fd9893"));
         executeTest("testMulti", spec);
     }
+
+    @Test
+    public void testComplex() {
+        WalkerTestSpec spec = new WalkerTestSpec(
+                baseTestString() + " -B set1,VCF," + validationDataLocation + "complexExample.vcf -B set2,VCF," + validationDataLocation + "complexExample.vcf -CT NWayVenn", 1,
+                Arrays.asList("8b72e557c0dd111738eaa69e9003fb3f"));
+        executeTest("testComplex", spec);
+    }
 }