From c028f13519a4f60ae41b9418e230a72552db39b3 Mon Sep 17 00:00:00 2001
From: Onur Sumer <s.onur.sumer@gmail.com>
Date: Fri, 25 Oct 2024 14:51:37 -0400
Subject: [PATCH] make clinical data sample/patient ids match legacy ids, and
 use stable ids to count samples when internal id is null

---
 .../impl/ClinicalDataDensityPlotServiceImpl.java     |  2 +-
 .../service/impl/ViolinPlotServiceImpl.java          | 12 ++++++++----
 .../columnar/util/ClinicalDataViolinPlotUtil.java    |  4 ++--
 .../mybatisclickhouse/StudyViewMapper.xml            |  6 +++---
 4 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/src/main/java/org/cbioportal/service/impl/ClinicalDataDensityPlotServiceImpl.java b/src/main/java/org/cbioportal/service/impl/ClinicalDataDensityPlotServiceImpl.java
index b6e4a55e679..10e4d588397 100644
--- a/src/main/java/org/cbioportal/service/impl/ClinicalDataDensityPlotServiceImpl.java
+++ b/src/main/java/org/cbioportal/service/impl/ClinicalDataDensityPlotServiceImpl.java
@@ -33,7 +33,7 @@ public DensityPlotData getDensityPlotData(List<ClinicalData> sampleClinicalData,
         result.setBins(new ArrayList<>());
 
         Map<String, List<ClinicalData>> clinicalDataGroupedBySampleId = sampleClinicalData.stream().
-            collect(Collectors.groupingBy(ClinicalData::getSampleId));
+            collect(Collectors.groupingBy(c -> c.getStudyId() + "_" + c.getSampleId()));
 
         List<ClinicalData> extractedXYClinicalData = clinicalDataGroupedBySampleId.entrySet().stream()
             .filter(entry -> entry.getValue().size() == 2 &&
diff --git a/src/main/java/org/cbioportal/service/impl/ViolinPlotServiceImpl.java b/src/main/java/org/cbioportal/service/impl/ViolinPlotServiceImpl.java
index 53e7f620874..dc12a2400ec 100644
--- a/src/main/java/org/cbioportal/service/impl/ViolinPlotServiceImpl.java
+++ b/src/main/java/org/cbioportal/service/impl/ViolinPlotServiceImpl.java
@@ -37,9 +37,11 @@ public ClinicalViolinPlotData getClinicalViolinPlotData(
         result.setRows(new ArrayList<>());
         
         // collect filtered samples into a set for quick lookup
-        Set<Integer> samplesForSampleCountsIds = 
+        Set<String> samplesForSampleCountsIds = 
             samplesForSampleCounts.stream()
-                .map(Sample::getInternalId)
+                .map(s -> s.getInternalId() == null ? 
+                    s.getCancerStudyIdentifier() + "_" + s.getStableId():  s.getInternalId().toString()
+                )
                 .collect(Collectors.toSet());
         
         // clinicalDataMap is a map sampleId->studyId->data
@@ -212,12 +214,14 @@ public ClinicalViolinPlotData getClinicalViolinPlotData(
     
     @SafeVarargs
     private static int countFilteredSamples(
-        Set<Integer> filteredSampleIds,
+        Set<String> filteredSampleIds,
         List<ClinicalData>... dataLists
     ) {
         return (int) Arrays.stream(dataLists)
             .flatMap(Collection::stream)
-            .map(ClinicalData::getInternalId)
+            .map(c -> c.getInternalId() == null ? 
+                c.getStudyId() + "_" + c.getSampleId() : c.getInternalId().toString()
+            )
             .filter(filteredSampleIds::contains)
             .count();
     }
diff --git a/src/main/java/org/cbioportal/web/columnar/util/ClinicalDataViolinPlotUtil.java b/src/main/java/org/cbioportal/web/columnar/util/ClinicalDataViolinPlotUtil.java
index 271d0745f25..597845db2ef 100644
--- a/src/main/java/org/cbioportal/web/columnar/util/ClinicalDataViolinPlotUtil.java
+++ b/src/main/java/org/cbioportal/web/columnar/util/ClinicalDataViolinPlotUtil.java
@@ -25,7 +25,7 @@ public static List<ClinicalData> convertPatientClinicalDataToSampleClinicalData(
         Map<String, Map<String, List<Sample>>> patientToSamples = samplesWithoutNumericalFilter
             .stream()
             .collect(Collectors.groupingBy(
-                s -> s.getCancerStudyIdentifier() + "_" + s.getPatientStableId(),
+                Sample::getPatientStableId,
                 Collectors.groupingBy(Sample::getCancerStudyIdentifier)
             ));
 
@@ -41,7 +41,7 @@ public static List<ClinicalData> convertPatientClinicalDataToSampleClinicalData(
                     newData.setPatientId(d.getPatientId());
                     newData.setStudyId(d.getStudyId());
                     newData.setAttrValue(d.getAttrValue());
-                    newData.setSampleId(s.getCancerStudyIdentifier() + "_" + s.getStableId());
+                    newData.setSampleId(s.getStableId());
 
                     sampleClinicalDataList.add(newData);
                 }
diff --git a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
index 7ed1c05f8d3..fd7d37b6d6f 100644
--- a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
+++ b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
@@ -91,8 +91,8 @@
 
     <select id="getSampleClinicalDataFromStudyViewFilter" resultType="org.cbioportal.model.ClinicalData">
         SELECT
-            sample_unique_id as sampleId,
-            patient_unique_id as patientId,
+            replaceOne(sample_unique_id, concat(cancer_study_identifier, '_'), '') as sampleId,
+            replaceOne(patient_unique_id, concat(cancer_study_identifier, '_'), '') as patientId,
             attribute_name as attrId,
             attribute_value as attrValue,
             cancer_study_identifier as studyId
@@ -113,7 +113,7 @@
 
     <select id="getPatientClinicalDataFromStudyViewFilter" resultType="org.cbioportal.model.ClinicalData">
         SELECT
-            patient_unique_id as patientId,
+            replaceOne(patient_unique_id, concat(cancer_study_identifier, '_'), '') as patientId,
             attribute_name as attrId,
             attribute_value as attrValue,
             cancer_study_identifier as studyId