Add bucket filtering optimization to genomic processing

hms-dbmi · Jan 9, 2024 · b241bed · b241bed
1 parent a63e6dd
commit b241bed
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 8 deletions.
diff --git a/data/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/data/genotype/BucketIndexBySample.java b/data/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/data/genotype/BucketIndexBySample.java
@@ -144,12 +144,12 @@ public BucketIndexBySample(VariantStore variantStore, String storageDir) throws
 	 * @return
 	 * @throws IOException 
 	 */
-	public Collection<String> filterVariantSetForPatientSet(Set<String> variantSet, List<Integer> patientSet) throws IOException{
-		
-		//a bitmask of which buckets contain any relevant variant. 
-		BigInteger patientBucketMask = patientSet.size() == 0 ? 
-				new BigInteger(new String(emptyBucketMaskChar()),2) : patientBucketMasks.get(patientSet.get(0));
-		
+	public Set<String> filterVariantSetForPatientSet(Set<String> variantSet, Collection<Integer> patientSet) throws IOException{
+
+		BigInteger patientBucketMask = patientSet.stream().findFirst()
+				.map(id -> patientBucketMasks.get(id))
+				.orElseGet(() -> new BigInteger(new String(emptyBucketMaskChar()),2));
+
 		BigInteger _defaultMask = patientBucketMask;
 		List<BigInteger> patientBucketmasksForSet = patientSet.parallelStream()
 				.map((patientNum)-> patientBucketMasks.get(patientNum))

diff --git a/...rc/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/PatientVariantJoinHandler.java b/...rc/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/PatientVariantJoinHandler.java
@@ -45,7 +45,10 @@ public BigInteger getPatientIdsForIntersectionOfVariantSets(Set<Integer> patient
 
             Set<String> variantsInScope = intersectionOfInfoFilters.mapToVariantSpec(variantService.getVariantIndex());
 
-            // todo: use BucketIndexBySample.filterVariantSetForPatientSet here?
+            // todo: determine ideal ratio to bother with this
+            if (patientsInScope.size() < variantService.getPatientIds().length) {
+                variantsInScope = variantService.filterVariantSetForPatientSet(variantsInScope, patientsInScope);
+            }
             Collection<List<String>> values = variantsInScope.stream()
                     .collect(Collectors.groupingByConcurrent((variantSpec) -> {
                         return new VariantSpec(variantSpec).metadata.offset / 1000;

diff --git a/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/VariantService.java b/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/VariantService.java
@@ -44,7 +44,7 @@ public String[] getVariantIndex() {
     public BucketIndexBySample getBucketIndex() {
         return bucketIndex;
     }
-    public Collection<String> filterVariantSetForPatientSet(Set<String> variantSet, List<Integer> patientSet) {
+    public Set<String> filterVariantSetForPatientSet(Set<String> variantSet, Collection<Integer> patientSet) {
         try {
             return bucketIndex.filterVariantSetForPatientSet(variantSet, patientSet);
         } catch (IOException e) {