Improve treatment data api performance

cBioPortal · Jul 10, 2023 · 719e5d2 · 719e5d2
1 parent 249d4ca
commit 719e5d2
Show file tree

Hide file tree

Showing 9 changed files with 45 additions and 40 deletions.
diff --git a/db-scripts/src/main/resources/cgds.sql b/db-scripts/src/main/resources/cgds.sql
@@ -755,5 +755,5 @@ CREATE TABLE `resource_study` (
 );
 
 -- THIS MUST BE KEPT IN SYNC WITH db.version PROPERTY IN pom.xml
-INSERT INTO info VALUES ('2.13.0', NULL);
+INSERT INTO info VALUES ('2.13.1', NULL);
 
diff --git a/db-scripts/src/main/resources/migration.sql b/db-scripts/src/main/resources/migration.sql
@@ -1021,3 +1021,9 @@ ALTER TABLE `mutation_event` CHANGE COLUMN `ONCOTATOR_PROTEIN_POS_START` `PROTEI
 ALTER TABLE `mutation_event` CHANGE COLUMN `ONCOTATOR_PROTEIN_POS_END` `PROTEIN_POS_END` int(11);
 UPDATE `info` SET `DB_SCHEMA_VERSION`="2.13.0";
 
+
+##version: 2.13.1
+ALTER TABLE `clinical_event_data` MODIFY COLUMN `VALUE` varchar(3000) NOT NULL;
+CREATE INDEX idx_clinical_event_key ON clinical_event_data (`KEY`);
+CREATE INDEX idx_clinical_event_value ON clinical_event_data (`VALUE`);
+UPDATE `info` SET `DB_SCHEMA_VERSION`="2.13.1";
diff --git a/...istence/persistence-api/src/main/java/org/cbioportal/persistence/TreatmentRepository.java b/...istence/persistence-api/src/main/java/org/cbioportal/persistence/TreatmentRepository.java
@@ -23,10 +23,10 @@ public interface TreatmentRepository {
     public Set<String> getAllUniqueTreatments(List<String> sampleIds, List<String> studyIds, ClinicalEventKeyCode key);
 
     @Cacheable(cacheResolver = "generalRepositoryCacheResolver", condition = "@cacheEnabledConfig.getEnabled()")
-    public Integer getTreatmentCount(List<String> studies, String key);
+    public Boolean hasTreatmentData(List<String> studies, String key);
 
     @Cacheable(cacheResolver = "generalRepositoryCacheResolver", condition = "@cacheEnabledConfig.getEnabled()")
-    public Integer getSampleCount(List<String> studies);
+    public Boolean hasSampleTimelineData(List<String> studies);
 
     @Cacheable(cacheResolver = "generalRepositoryCacheResolver", condition = "@cacheEnabledConfig.getEnabled()")
     public Boolean studyIdHasTreatments(String studyId, ClinicalEventKeyCode key);

diff --git a/...persistence-mybatis/src/main/java/org/cbioportal/persistence/mybatis/TreatmentMapper.java b/...persistence-mybatis/src/main/java/org/cbioportal/persistence/mybatis/TreatmentMapper.java
@@ -15,9 +15,9 @@ public interface TreatmentMapper {
 
     Set<String> getAllUniqueTreatments(List<String> sampleIds, List<String> studyIds, String key);
 
-    Integer getTreatmentCount(List<String> sampleIds, List<String> studyIds, String key);
+    Boolean hasTreatmentData(List<String> sampleIds, List<String> studyIds, String key);
 
-    Integer getSampleCount(List<String> sampleIds, List<String> studyIds);
+    Boolean hasSampleTimelineData(List<String> sampleIds, List<String> studyIds);
 
     Boolean studyIdHasTreatments(String studyId, String key);
 }
diff --git a/...-mybatis/src/main/java/org/cbioportal/persistence/mybatis/TreatmentMyBatisRepository.java b/...-mybatis/src/main/java/org/cbioportal/persistence/mybatis/TreatmentMyBatisRepository.java
@@ -75,13 +75,13 @@ public Set<String> getAllUniqueTreatments(List<String> sampleIds, List<String> s
     }
 
     @Override
-    public Integer getTreatmentCount(List<String> studies, String key) {
-        return treatmentMapper.getTreatmentCount(null, studies, key);
+    public Boolean hasTreatmentData(List<String> studies, String key) {
+        return treatmentMapper.hasTreatmentData(null, studies, key);
     }
 
     @Override
-    public Integer getSampleCount(List<String> studies) {
-        return treatmentMapper.getSampleCount(null, studies);
+    public Boolean hasSampleTimelineData(List<String> studies) {
+        return treatmentMapper.hasSampleTimelineData(null, studies);
     }
 
     @Override

diff --git a/...istence-mybatis/src/main/resources/org/cbioportal/persistence/mybatis/TreatmentMapper.xml b/...istence-mybatis/src/main/resources/org/cbioportal/persistence/mybatis/TreatmentMapper.xml
@@ -61,22 +61,23 @@
             AND clinical_event_data.KEY = #{key}
     </select>
 
-    <select id="getTreatmentCount" resultType="java.lang.Integer">
-        SELECT
-            COUNT(clinical_event.PATIENT_ID)
+    <select id="hasTreatmentData" resultType="java.lang.Boolean">
+        SELECT EXISTS(SELECT
+            *
         FROM
             clinical_event
             INNER JOIN clinical_event_data ON clinical_event.CLINICAL_EVENT_ID = clinical_event_data.CLINICAL_EVENT_ID
             INNER JOIN patient ON clinical_event.PATIENT_ID = patient.INTERNAL_ID
             INNER JOIN sample ON patient.INTERNAL_ID = sample.PATIENT_ID
             INNER JOIN cancer_study ON patient.CANCER_STUDY_ID = cancer_study.CANCER_STUDY_ID
         <include refid="where"/>
-            AND clinical_event_data.KEY = #{key}
+            AND clinical_event_data.KEY = #{key} LIMIT 1
+        )
     </select>
 
-    <select id="getSampleCount" resultType="java.lang.Integer">
-        SELECT
-            COUNT(sample.STABLE_ID)
+    <select id="hasSampleTimelineData" resultType="java.lang.Boolean">
+        SELECT EXISTS(SELECT
+            *
         FROM
             clinical_event
             INNER JOIN clinical_event_data ON clinical_event.CLINICAL_EVENT_ID = clinical_event_data.CLINICAL_EVENT_ID
@@ -85,20 +86,21 @@
             INNER JOIN cancer_study ON patient.CANCER_STUDY_ID = cancer_study.CANCER_STUDY_ID
         <include refid="where"/>
             AND clinical_event_data.KEY = 'SAMPLE_ID'
-            AND (clinical_event.EVENT_TYPE LIKE 'Sample Acquisition' OR clinical_event.EVENT_TYPE LIKE 'SPECIMEN')
+            AND (clinical_event.EVENT_TYPE LIKE 'Sample Acquisition' OR clinical_event.EVENT_TYPE LIKE 'SPECIMEN') LIMIT 1)
     </select>
 
+
     <select id="studyIdHasTreatments" resultType="java.lang.Boolean">
-        SELECT
-            count(cancer_study.CANCER_STUDY_IDENTIFIER) > 0
+        SELECT EXISTS(SELECT
+            *
         FROM
             clinical_event
             INNER JOIN clinical_event_data ON clinical_event.CLINICAL_EVENT_ID = clinical_event_data.CLINICAL_EVENT_ID
             INNER JOIN patient ON clinical_event.PATIENT_ID = patient.INTERNAL_ID
             INNER JOIN cancer_study ON patient.CANCER_STUDY_ID = cancer_study.CANCER_STUDY_ID
         WHERE
             clinical_event_data.KEY = #{key}
-            AND cancer_study.CANCER_STUDY_IDENTIFIER = #{studyId}
+            AND cancer_study.CANCER_STUDY_IDENTIFIER = #{studyId} LIMIT 1)
     </select>
 
     <sql id="where">

diff --git a/...atis/src/test/java/org/cbioportal/persistence/mybatis/TreatmentMyBatisRepositoryTest.java b/...atis/src/test/java/org/cbioportal/persistence/mybatis/TreatmentMyBatisRepositoryTest.java
@@ -147,19 +147,17 @@ public void getAllUniqueTreatments() {
     }
 
     @Test
-    public void getTreatmentCount() {
-        Integer expected = 1;
-        Integer actual = treatmentRepository.getTreatmentCount(Collections.singletonList("study_tcga_pub"), ClinicalEventKeyCode.Agent.getKey());
+    public void hasTreatmentData() {
+        Boolean actual = treatmentRepository.hasTreatmentData(Collections.singletonList("study_tcga_pub"), ClinicalEventKeyCode.Agent.getKey());
 
-        Assert.assertEquals(actual, expected);
+        Assert.assertEquals(actual, true);
     }
 
     @Test
     public void getSampleCount() {
-        Integer expected = 2;
-        Integer actual = treatmentRepository.getSampleCount(Collections.singletonList("study_tcga_pub"));
+        Boolean actual = treatmentRepository.hasSampleTimelineData(Collections.singletonList("study_tcga_pub"));
 
-        Assert.assertEquals(actual, expected);
+        Assert.assertEquals(actual, true);
     }
 
 }
diff --git a/pom.xml b/pom.xml
@@ -314,7 +314,7 @@
     <tomcat.session.timeout>720</tomcat.session.timeout>
 
     <!-- THIS SHOULD BE KEPT IN SYNC TO VERSION IN CGDS.SQL -->
-    <db.version>2.13.0</db.version>
+    <db.version>2.13.1</db.version>
   </properties>
 
   <modules>

diff --git a/service/src/main/java/org/cbioportal/service/impl/TreatmentServiceImpl.java b/service/src/main/java/org/cbioportal/service/impl/TreatmentServiceImpl.java
@@ -12,7 +12,7 @@
 import java.util.stream.Collectors;
 import java.util.stream.Stream;
 
-import static java.util.stream.Collectors.toList;
+import static java.util.stream.Collectors.*;
 
 @Service
 public class TreatmentServiceImpl implements TreatmentService {
@@ -40,16 +40,16 @@ private Pair<List<String>, List<String>> filterIds(List<String> sampleIds, List<
         }
         return new ImmutablePair<>(filteredSampleIds, filteredStudyIds);
     }
-    
+
     @Override
     public List<SampleTreatmentRow> getAllSampleTreatmentRows(List<String> sampleIds, List<String> studyIds, ClinicalEventKeyCode key) {
         Pair<List<String>, List<String>> filteredIds = filterIds(sampleIds, studyIds, key);
         sampleIds = filteredIds.getLeft();
         studyIds = filteredIds.getRight();
 
-        Map<String, List<ClinicalEventSample>> samplesByPatient = 
+        Map<String, List<ClinicalEventSample>> samplesByPatient =
             treatmentRepository.getSamplesByPatientId(sampleIds, studyIds);
-        Map<String, List<Treatment>> treatmentsByPatient = 
+        Map<String, List<Treatment>> treatmentsByPatient =
             treatmentRepository.getTreatmentsByPatientId(sampleIds, studyIds, key);
 
         Stream<SampleTreatmentRow> rows = samplesByPatient.keySet().stream()
@@ -150,16 +150,15 @@ Stream<SampleTreatmentRow> toRows() {
             );
         }
     }
-
     @Override
     public List<PatientTreatmentRow> getAllPatientTreatmentRows(
         List<String> sampleIds, List<String> studyIds, ClinicalEventKeyCode key
     ) {
         Pair<List<String>, List<String>> filteredIds = filterIds(sampleIds, studyIds, key);
         sampleIds = filteredIds.getLeft();
         studyIds = filteredIds.getRight();
-        
-        Map<String, List<Treatment>> treatmentsByPatient = 
+
+        Map<String, List<Treatment>> treatmentsByPatient =
             treatmentRepository.getTreatmentsByPatientId(sampleIds, studyIds, key);
         Map<String, List<ClinicalEventSample>> samplesByPatient = treatmentRepository
             .getShallowSamplesByPatientId(sampleIds, studyIds)
@@ -204,17 +203,17 @@ private List<Map.Entry<String, List<Treatment>>> matchingPatients(
 
     @Override
     public Boolean containsTreatmentData(List<String> studies, ClinicalEventKeyCode key) {
-        return treatmentRepository.getTreatmentCount(studies, key.getKey()) > 0;
+        return treatmentRepository.hasTreatmentData(studies, key.getKey());
     }
 
     @Override
     public Boolean containsSampleTreatmentData(List<String> studyIds, ClinicalEventKeyCode key) {
         studyIds = studyIds.stream()
             .filter(studyId -> treatmentRepository.studyIdHasTreatments(studyId, key))
             .collect(Collectors.toList());
-        Integer sampleCount = treatmentRepository.getSampleCount(studyIds);
-        Integer treatmentCount = treatmentRepository.getTreatmentCount(studyIds, key.getKey());
-
-        return sampleCount * treatmentCount > 0;
+        if(studyIds.size() > 0 && treatmentRepository.hasSampleTimelineData(studyIds)) {
+            return treatmentRepository.hasTreatmentData(studyIds, key.getKey());
+        }
+        return false;
     }
 }