From 454bf20298e8722176c285aef2401dd1b954ea3f Mon Sep 17 00:00:00 2001
From: "Zhaoyuan (Ryan) Fu" <fuzhaoyuan@gmail.com>
Date: Fri, 21 Jun 2024 10:56:12 -0400
Subject: [PATCH] Put calculation back into SQL for simpler code

---
 .../impl/StudyViewColumnarServiceImpl.java     | 11 +++--------
 .../mybatisclickhouse/StudyViewMapper.xml      | 18 +++++++++---------
 2 files changed, 12 insertions(+), 17 deletions(-)

diff --git a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
index cf90ea28bba..7f136b0e60e 100644
--- a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
+++ b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
@@ -129,16 +129,11 @@ public List<GenomicDataCountItem> getMutationCountsByGeneSpecific(StudyViewFilte
         // Error handling may need to be added
         for (GenomicDataFilter genomicDataFilter : genomicDataFilters) {
             Map<String, Integer> counts = studyViewRepository.getMutationCounts(studyViewFilter, categorizedClinicalDataCountFilter, genomicDataFilter);
-            int totalCount = counts.getOrDefault("totalCount", 0);
-            int profiledCount = counts.getOrDefault("profiledCount", 0);
-            int mutatedCount = counts.getOrDefault("mutatedCount", 0);
-            int notMutatedCount = profiledCount - mutatedCount;
-            int notProfiledCount = totalCount - profiledCount;
 
             List<GenomicDataCount> genomicDataCountList = List.of(
-                new GenomicDataCount("Mutated", "MUTATED", mutatedCount, mutatedCount),
-                new GenomicDataCount("Not Mutated", "NOT_MUTATED", notMutatedCount, notMutatedCount),
-                new GenomicDataCount("Not Profiled", "NOT_PROFILED", notProfiledCount, notProfiledCount)
+                new GenomicDataCount("Mutated", "MUTATED", counts.get("mutatedCount"), counts.get("mutatedCount")),
+                new GenomicDataCount("Not Mutated", "NOT_MUTATED", counts.get("notMutatedCount"), counts.get("notMutatedCount")),
+                new GenomicDataCount("Not Profiled", "NOT_PROFILED", counts.get("notProfiledCount"), counts.get("notProfiledCount"))
             );
 
             genomicDataCountItemList.add(new GenomicDataCountItem(genomicDataFilter.getHugoGeneSymbol(), "mutations", genomicDataCountList));
diff --git a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
index 8e16975111e..5d8a03d2cb0 100644
--- a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
+++ b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
@@ -219,21 +219,21 @@
 
     <!-- for /mutation-data-counts/fetch (returns GenomicDataCountItem objects) mutation counts pie chart part -->
     <select id="getMutationCounts">
-        WITH total_samples as (
-            SELECT DISTINCT sample_unique_id
+        WITH total_count as (
+            SELECT count(distinct sample_unique_id)
             FROM sample_to_gene_panel
             WHERE sample_unique_id IN (<include refid="sampleUniqueIdsFromStudyViewFilter"/>)
         ),
-        profiled_samples as (
-            SELECT DISTINCT stgp.sample_unique_id
+        profiled_count as (
+            SELECT count(distinct stgp.sample_unique_id)
             FROM sample_to_gene_panel stgp
                 JOIN gene_panel_to_gene gpg ON stgp.gene_panel_id = gpg.gene_panel_id
             WHERE
                 sample_unique_id IN (<include refid="sampleUniqueIdsFromStudyViewFilter"/>)
                 AND gpg.gene = #{genomicDataFilter.hugoGeneSymbol}
         ),
-        mutated_samples as (
-            SELECT DISTINCT sample_unique_id
+        mutated_count as (
+            SELECT count(distinct sample_unique_id)
             FROM genomic_event_derived
             WHERE
                 sample_unique_id IN (<include refid="sampleUniqueIdsFromStudyViewFilter"/>)
@@ -241,9 +241,9 @@
                 AND variant_type = 'mutation'
         )
         SELECT
-            cast((SELECT count(*) FROM total_samples) as INTEGER) as totalCount,
-            cast((SELECT count(*) FROM profiled_samples) as INTEGER) as profiledCount,
-            cast((SELECT count(*) FROM mutated_samples) as INTEGER) as mutatedCount
+            cast((SELECT * FROM mutated_count) as INTEGER) as mutatedCount,
+            cast(((SELECT * FROM profiled_count) - (SELECT * FROM mutated_count)) as INTEGER) as notMutatedCount,
+            cast(((SELECT * FROM total_count) - (SELECT * FROM profiled_count)) as INTEGER) as notProfiledCount
     </select>
 
     <!-- for /mutation-data-counts/fetch - (returns GenomicDataCountItem objects) mutation type counts table part-->