From 292e66a2d3a50923a899bc4c61ab500ec9a2e8eb Mon Sep 17 00:00:00 2001
From: ramari16 <ramari16@gmail.com>
Date: Tue, 22 Oct 2024 10:23:36 -0400
Subject: [PATCH] Fix bucket index loading for non-variant explorer
 environments (#123)

---
 .../processing/GenomicProcessorNodeImpl.java   |  5 +----
 .../hpds/processing/VariantService.java        | 18 +++++++++++++-----
 2 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/GenomicProcessorNodeImpl.java b/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/GenomicProcessorNodeImpl.java
index 68e0140b..ea612b89 100644
--- a/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/GenomicProcessorNodeImpl.java
+++ b/processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/GenomicProcessorNodeImpl.java
@@ -35,8 +35,6 @@ public class GenomicProcessorNodeImpl implements GenomicProcessor {
 
     private final VariantService variantService;
 
-    private final VariantMetadataIndex variantMetadataIndex;
-
 
     private final String HOMOZYGOUS_VARIANT = "1/1";
     private final String HETEROZYGOUS_VARIANT = "0/1";
@@ -46,7 +44,6 @@ public GenomicProcessorNodeImpl(String genomicDataDirectory) {
         this.genomicDataDirectory = genomicDataDirectory;
         this.variantService = new VariantService(genomicDataDirectory);
         this.patientVariantJoinHandler = new PatientVariantJoinHandler(variantService);
-        this.variantMetadataIndex = VariantMetadataIndex.createInstance(genomicDataDirectory);
 
         infoStores = new HashMap<>();
         File genomicDataDirectoryFile = new File(this.genomicDataDirectory);
@@ -402,6 +399,6 @@ public List<InfoColumnMeta> getInfoColumnMeta() {
 
     @Override
     public Map<String, Set<String>> getVariantMetadata(Collection<String> variantList) {
-        return variantMetadataIndex.findByMultipleVariantSpec(variantList);
+        return variantService.findByMultipleVariantSpec(variantList);
     }
 }
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
index 0981a6d1..c8f94850 100644
--- 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
@@ -1,9 +1,6 @@
 package edu.harvard.hms.dbmi.avillach.hpds.processing;
 
-import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.BucketIndexBySample;
-import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.VariableVariantMasks;
-import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.VariantMasks;
-import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.VariantStore;
+import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.*;
 import edu.harvard.hms.dbmi.avillach.hpds.data.genotype.caching.VariantBucketHolder;
 import edu.harvard.hms.dbmi.avillach.hpds.storage.FileBackedByteIndexedStorage;
 import edu.harvard.hms.dbmi.avillach.hpds.storage.FileBackedJavaIndexedStorage;
@@ -38,6 +35,8 @@ public class VariantService {
     private String[] variantIndex = null;
     private BucketIndexBySample bucketIndex;
 
+    private final VariantMetadataIndex variantMetadataIndex;
+
     public String[] getVariantIndex() {
         return variantIndex;
     }
@@ -60,6 +59,8 @@ public VariantService(String genomicDataDirectory) {
         BUCKET_INDEX_BY_SAMPLE_FILE = genomicDataDirectory + "BucketIndexBySample.javabin";
 
         variantStore = loadVariantStore();
+
+        this.variantMetadataIndex = VariantMetadataIndex.createInstance(genomicDataDirectory);
         try {
             loadGenomicCacheFiles();
         } catch (Exception e) {
@@ -162,7 +163,10 @@ private void loadGenomicCacheFiles() throws FileNotFoundException, IOException,
                     log.info("Found " + variantIndex.length + " total variants.");
                 }
             }
-            if(variantStore.getPatientIds().length > 0 && !new File(BUCKET_INDEX_BY_SAMPLE_FILE).exists()) {
+            // todo: not loading bucket index when there is no variant metadata index is a temporary fix for non-variant explorer environments
+            // once we start building the bucket index as part of the ETL, we can remove this check and leverage the bucket index
+            // for all genomic queries
+            if(variantStore.getPatientIds().length > 0 && variantMetadataIndex != null && !new File(BUCKET_INDEX_BY_SAMPLE_FILE).exists()) {
                 log.info("creating new " + BUCKET_INDEX_BY_SAMPLE_FILE);
                 bucketIndex = new BucketIndexBySample(variantStore, genomicDataDirectory);
                 try (
@@ -204,4 +208,8 @@ public List<VariableVariantMasks> getMasksForDbSnpSpec(String variantName) {
     public BigInteger emptyBitmask() {
         return variantStore.emptyBitmask();
     }
+
+    public Map<String, Set<String>> findByMultipleVariantSpec(Collection<String> variantList) {
+        return variantMetadataIndex.findByMultipleVariantSpec(variantList);
+    }
 }