optimise endpoint

thehyve · Dec 11, 2024 · bdc3c8f · bdc3c8f
1 parent 2b2868c
commit bdc3c8f
Show file tree

Hide file tree

Showing 4 changed files with 61 additions and 10 deletions.
diff --git a/dev/add_single_cell_table.sql b/dev/add_single_cell_table.sql
@@ -10,3 +10,8 @@ CREATE TABLE IF NOT EXISTS single_cell_expression (
     FOREIGN KEY(SAMPLE_ID) REFERENCES sample(INTERNAL_ID) ON DELETE CASCADE,
     FOREIGN KEY(ENTREZ_GENE_ID) REFERENCES gene(ENTREZ_GENE_ID) ON DELETE CASCADE
 );
+
+CREATE INDEX single_cell_sortby ON single_cell_expression (GENETIC_PROFILE_ID, SAMPLE_ID, TISSUE, CELL_TYPE);
+CREATE INDEX sample_index ON single_cell_expression (GENETIC_PROFILE_ID, SAMPLE_ID);
+CREATE INDEX entrez_index ON single_cell_expression (ENTREZ_GENE_ID);
+
diff --git a/dev/create_sc_expression_sql.py b/dev/create_sc_expression_sql.py
@@ -25,6 +25,11 @@
     FOREIGN KEY(SAMPLE_ID) REFERENCES sample(INTERNAL_ID),
     FOREIGN KEY(ENTREZ_GENE_ID) REFERENCES gene(ENTREZ_GENE_ID)
 );
+
+
+CREATE INDEX single_cell_sortby ON single_cell_expression (GENETIC_PROFILE_ID, SAMPLE_ID, TISSUE, CELL_TYPE);
+CREATE INDEX sample_index ON single_cell_expression (GENETIC_PROFILE_ID, SAMPLE_ID);
+CREATE INDEX entrez_index ON single_cell_expression (ENTREZ_GENE_ID);
 """
 
 add_genetic_profile = f"""

diff --git a/src/main/java/org/cbioportal/model/SingleCellExpressionGroupedBySample.java b/src/main/java/org/cbioportal/model/SingleCellExpressionGroupedBySample.java
@@ -1,5 +1,7 @@
 package org.cbioportal.model;
 
+import java.util.LinkedHashMap;
+import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
 import java.util.stream.Collectors;
@@ -26,8 +28,8 @@ public SingleCellExpressionGroupedBySample(String sampleId, List<SingleCellExpre
         this.patientId = singleCellExpressions.get(0).getPatientId();
         this.studyId = singleCellExpressions.get(0).getStudyId();
 
-        Map<String, List<SingleCellExpression>> tissueMap = singleCellExpressions.stream()
-            .collect(Collectors.groupingBy(SingleCellExpression::getTissue));
+        Map<String, List<SingleCellExpression>> tissueMap = SingleCellExpressionGroupedBySample
+            .createSingleCellExpressionGroupedMap(singleCellExpressions, "tissue");
         this.singleCellExpressionGroupedByTissues = tissueMap.entrySet().stream()
             .map(e -> new SingleCellExpressionGroupedByTissue(e.getKey(), e.getValue()))
             .collect(Collectors.toList());
@@ -75,8 +77,9 @@ public SingleCellExpressionGroupedByTissue(String tissue,
                 List<SingleCellExpression> singleCellExpressions) {
             this.tissue = tissue;
 
-            Map<String, List<SingleCellExpression>> cellTypeMap = singleCellExpressions.stream()
-                .collect(Collectors.groupingBy(SingleCellExpression::getCellType));
+            Map<String, List<SingleCellExpression>> cellTypeMap = SingleCellExpressionGroupedBySample
+            .createSingleCellExpressionGroupedMap(singleCellExpressions, "cellType");
+
             this.singleCellExpressionGroupedByCellTypes = cellTypeMap.entrySet().stream()
                 .map(e -> new SingleCellExpressionGroupedByCellType(e.getKey(), e.getValue()))
                 .collect(Collectors.toList());
@@ -147,4 +150,43 @@ public void setExpressionValue(Float expressionValue) {
             }
         }
     }
-}
+
+    public static Map<String, List<SingleCellExpression>> createSingleCellExpressionGroupedMap(List<SingleCellExpression> singleCellExpressionData, String sortBy) {
+        // singleCellExpressionData NEEDS TO BE SORTED in order for this to work correctly.
+        // This is done this way in order to improve efficiency. Sorting is done in the DB query,
+        // and indices on the relevant columns increase query speed
+        Map<String, List<SingleCellExpression>> singleCellExpressionGroupedBy = new LinkedHashMap<>();
+        String currentValue = "";
+        String forLoopValue = "";
+        List<SingleCellExpression> singleCellExpressionList = new LinkedList<>();
+        for (SingleCellExpression singleCellExpression : singleCellExpressionData) {
+            switch (sortBy) {
+                case "sampleId":
+                    forLoopValue = singleCellExpression.getSampleId();
+                    break;
+                case "tissue":
+                    forLoopValue = singleCellExpression.getTissue();
+                    break;
+                case "cellType":
+                    forLoopValue = singleCellExpression.getCellType();
+                    break;
+                default:
+                    throw new RuntimeException("sortBy option " + sortBy + " not supported.");
+            }
+            if (currentValue.equals("")) {
+                currentValue = forLoopValue;
+                singleCellExpressionList = new LinkedList<>();
+            }
+            if (!currentValue.equals(forLoopValue)) {
+                singleCellExpressionGroupedBy.put(currentValue, singleCellExpressionList);
+                currentValue = forLoopValue;
+                singleCellExpressionList = new LinkedList<>();
+            }
+            singleCellExpressionList.add(singleCellExpression);
+        }
+        if (!currentValue.equals("")) {
+            singleCellExpressionGroupedBy.put(currentValue, singleCellExpressionList);
+        }
+        return singleCellExpressionGroupedBy;
+    }
+}
diff --git a/src/main/java/org/cbioportal/service/impl/SingleCellExpressionServiceImpl.java b/src/main/java/org/cbioportal/service/impl/SingleCellExpressionServiceImpl.java
@@ -8,7 +8,6 @@
 import org.cbioportal.service.exception.MolecularProfileNotFoundException;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.stereotype.Service;
-
 import org.cbioportal.model.SingleCellExpression;
 import org.cbioportal.model.SingleCellExpressionGroupedBySample;
 import org.cbioportal.persistence.SingleCellExpressionRepository;
@@ -37,14 +36,14 @@ public List<SingleCellExpressionGroupedBySample> fetchSingleCellExpressionInMole
             throws MolecularProfileNotFoundException {
 
         List<SingleCellExpression> singleCellExpressionData = fetchSingleCellExpressionInMolecularProfile(
-                molecularProfileId, sampleListId, entrezGeneIds, pageSize, pageNumber, "sampleId", "ASC");
+                molecularProfileId, sampleListId, entrezGeneIds, pageSize, pageNumber, "sampleId ASC, tissue ASC, cellType", "ASC");
 
-        Map<String, List<SingleCellExpression>> singleCellExpressionGroupedBySamples = singleCellExpressionData.stream()
-                .collect(Collectors.groupingBy(SingleCellExpression::getSampleId));
+        Map<String, List<SingleCellExpression>> singleCellExpressionGroupedBySamples = 
+            SingleCellExpressionGroupedBySample.createSingleCellExpressionGroupedMap(singleCellExpressionData, "sampleId");
 
         return singleCellExpressionGroupedBySamples.entrySet().stream()
                 .map(e -> new SingleCellExpressionGroupedBySample(e.getKey(), e.getValue()))
                 .collect(Collectors.toList());
     }
 
-}
+}