From 7cf0368ba1918629f8c261c08978fa8ae178e455 Mon Sep 17 00:00:00 2001
From: "Zhaoyuan (Ryan) Fu" <fuzhaoyuan@gmail.com>
Date: Mon, 2 Dec 2024 10:36:50 -0500
Subject: [PATCH] Merge genomic data bins working

---
 .../impl/StudyViewColumnarServiceImpl.java    |  1 +
 .../web/parameter/DataFilterValue.java        |  8 ++++
 .../web/parameter/StudyViewFilter.java        | 45 +++++++++++++++++++
 3 files changed, 54 insertions(+)

diff --git a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
index 7063f64f560..422401e62af 100644
--- a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
+++ b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
@@ -285,6 +285,7 @@ public List<GenomicDataCountItem> getMutationTypeCountsByGeneSpecific(StudyViewF
     
     
     private StudyViewFilterContext createContext(StudyViewFilter studyViewFilter) {
+        studyViewFilter.mergeDataFilterNumericalValues();
         List<CustomSampleIdentifier> customSampleIdentifiers = customDataFilterUtil.extractCustomDataSamples(studyViewFilter);
         return new StudyViewFilterContext(studyViewFilter, customSampleIdentifiers);
     }
diff --git a/src/main/java/org/cbioportal/web/parameter/DataFilterValue.java b/src/main/java/org/cbioportal/web/parameter/DataFilterValue.java
index 184aee40f29..af434cfa55d 100644
--- a/src/main/java/org/cbioportal/web/parameter/DataFilterValue.java
+++ b/src/main/java/org/cbioportal/web/parameter/DataFilterValue.java
@@ -9,6 +9,14 @@ public class DataFilterValue implements Serializable {
     private BigDecimal end;
     private String value;
 
+    public DataFilterValue() {}
+    
+    public DataFilterValue(BigDecimal start, BigDecimal end, String value) {
+        this.start = start;
+        this.end = end;
+        this.value = value;
+    }
+    
     public BigDecimal getStart() {
         return start;
     }
diff --git a/src/main/java/org/cbioportal/web/parameter/StudyViewFilter.java b/src/main/java/org/cbioportal/web/parameter/StudyViewFilter.java
index 738d3632a70..95ae884bddd 100644
--- a/src/main/java/org/cbioportal/web/parameter/StudyViewFilter.java
+++ b/src/main/java/org/cbioportal/web/parameter/StudyViewFilter.java
@@ -1,6 +1,8 @@
 package org.cbioportal.web.parameter;
 
 import java.io.Serializable;
+import java.math.BigDecimal;
+import java.util.ArrayList;
 import java.util.List;
 import java.util.Objects;
 
@@ -41,6 +43,7 @@ public class StudyViewFilter implements Serializable {
     private AlterationFilter alterationFilter;
     private List<DataFilter> clinicalEventFilters;
     private List<MutationDataFilter> mutationDataFilters;
+    private static boolean areBinsMerged = false;
     
     @AssertTrue
     private boolean isEitherSampleIdentifiersOrStudyIdsPresent() {
@@ -230,4 +233,46 @@ public void setClinicalEventFilters(List<DataFilter> clinicalEventFilters) {
     public void setMutationDataFilters(List<MutationDataFilter> mutationDataFilters) {
         this.mutationDataFilters = mutationDataFilters;
     }
+    
+    /**
+     * Merge the range of numerical values in DataFilters to reduce the number of scans that runs on the database.
+     * Variable 'areBinsMerged' is static so this method only gets run once.
+     */
+    public void mergeDataFilterNumericalValues() {
+        if (areBinsMerged || this.genomicDataFilters == null || this.genomicDataFilters.isEmpty()) return;
+        
+        List<GenomicDataFilter> mergedGenomicDataFilters = new ArrayList<>();
+        
+        for (GenomicDataFilter genomicDataFilter : this.genomicDataFilters) {
+            GenomicDataFilter mergedGenomicDataFilter = new GenomicDataFilter(genomicDataFilter.getHugoGeneSymbol(), genomicDataFilter.getProfileType());
+            List<DataFilterValue> mergedValues = new ArrayList<>();
+            
+            boolean hasNullStart = false, hasNullEnd = false;
+            BigDecimal mergedStart = null, mergedEnd = null;
+            for (DataFilterValue dataFilterValue : genomicDataFilter.getValues()) {
+                // filter non-numerical values and keep them intact
+                if (dataFilterValue.getValue() != null) {
+                    mergedValues.add(dataFilterValue);
+                }
+                // record if numerical values have null start or end, otherwise record their start-end range
+                else {
+                    if (dataFilterValue.getStart() == null) hasNullStart = true;
+                    else if (mergedStart == null) mergedStart = dataFilterValue.getStart();
+                    else if (dataFilterValue.getStart().compareTo(mergedStart) < 0) mergedStart = dataFilterValue.getStart();
+                    if (dataFilterValue.getEnd() == null) hasNullEnd = true;
+                    else if (mergedEnd == null) mergedEnd = dataFilterValue.getEnd();
+                    else if (dataFilterValue.getEnd().compareTo(mergedEnd) > 0) mergedEnd = dataFilterValue.getEnd();
+                }
+            }
+            if (hasNullStart) mergedStart = null;
+            if (hasNullEnd) mergedEnd = null;
+            
+            mergedValues.add(new DataFilterValue(mergedStart, mergedEnd, null));
+            mergedGenomicDataFilter.setValues(mergedValues);
+            mergedGenomicDataFilters.add(mergedGenomicDataFilter);
+        }
+        
+        this.genomicDataFilters = mergedGenomicDataFilters;
+        areBinsMerged = true;
+    }
 }