Merge pull request #626 from julie-sullivan/cnvpadding

CellBase v4 - fix cnvPadding bug
opencb · Nov 14, 2022 · d01e2ab · d01e2ab
2 parents b96d74b + 4a4af00
commit d01e2ab
Show file tree

Hide file tree

Showing 11 changed files with 54 additions and 16 deletions.
diff --git a/cellbase-app/pom.xml b/cellbase-app/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <groupId>org.opencb.cellbase</groupId>
         <artifactId>cellbase</artifactId>
-        <version>4.11.1</version>
+        <version>4.12.0-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

diff --git a/cellbase-client/pom.xml b/cellbase-client/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <groupId>org.opencb.cellbase</groupId>
         <artifactId>cellbase</artifactId>
-        <version>4.11.1</version>
+        <version>4.12.0-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

diff --git a/cellbase-core/pom.xml b/cellbase-core/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <groupId>org.opencb.cellbase</groupId>
         <artifactId>cellbase</artifactId>
-        <version>4.11.1</version>
+        <version>4.12.0-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

diff --git a/.../src/main/java/org/opencb/cellbase/core/variant/annotation/ConsequenceTypeCalculator.java b/.../src/main/java/org/opencb/cellbase/core/variant/annotation/ConsequenceTypeCalculator.java
@@ -8,6 +8,7 @@
 import org.opencb.biodata.models.variant.avro.ConsequenceType;
 import org.opencb.biodata.models.variant.avro.ProteinVariantAnnotation;
 import org.opencb.biodata.models.variant.avro.SequenceOntologyTerm;
+import org.opencb.biodata.models.variant.avro.VariantType;
 import org.opencb.cellbase.core.api.GenomeDBAdaptor;
 import org.opencb.commons.datastore.core.Query;
 import org.opencb.commons.datastore.core.QueryOptions;
@@ -53,7 +54,9 @@ protected void parseQueryParam(QueryOptions queryOptions) {
     }
 
     protected int getStart(int extraPadding) {
-        if (imprecise && variant.getSv() != null) {
+        if (imprecise && VariantType.CNV.equals(variant.getType())) {
+            return variant.getStart() - extraPadding;
+        } else if (imprecise && variant.getSv() != null) {
             return variant.getSv().getCiStartLeft() != null ? variant.getSv().getCiStartLeft() - extraPadding
                     : variant.getStart();
         } else {
@@ -63,7 +66,9 @@ protected int getStart(int extraPadding) {
     }
 
     protected int getEnd(int extraPadding) {
-        if (imprecise && variant.getSv() != null) {
+        if (imprecise && VariantType.CNV.equals(variant.getType())) {
+            return variant.getEnd() + extraPadding;
+        } else if (imprecise && variant.getSv() != null) {
             return variant.getSv().getCiEndRight() != null ? variant.getSv().getCiEndRight() + extraPadding
                     : variant.getEnd();
         } else {

diff --git a/...a/org/opencb/cellbase/core/variant/annotation/ConsequenceTypeGenericRegionCalculator.java b/...a/org/opencb/cellbase/core/variant/annotation/ConsequenceTypeGenericRegionCalculator.java
@@ -6,6 +6,7 @@
 import org.opencb.biodata.models.variant.Variant;
 import org.opencb.biodata.models.variant.avro.ConsequenceType;
 import org.opencb.biodata.models.variant.avro.ExonOverlap;
+import org.opencb.biodata.models.variant.avro.VariantType;
 import org.opencb.commons.datastore.core.QueryOptions;
 
 import java.util.ArrayList;
@@ -24,8 +25,9 @@ public List<ConsequenceType> run(Variant inputVariant, List<Gene> geneList, bool
         parseQueryParam(queryOptions);
         List<ConsequenceType> consequenceTypeList = new ArrayList<>();
         variant = inputVariant;
-        variantEnd = getEnd(svExtraPadding);
-        variantStart = getStart(svExtraPadding);
+        int extraPadding = VariantType.CNV.equals(variant.getType()) ? cnvExtraPadding : svExtraPadding;
+        variantEnd = getEnd(extraPadding);
+        variantStart = getStart(extraPadding);
 //        isBigDeletion = ((variantEnd - variantStart) > BIG_VARIANT_SIZE_THRESHOLD);
         boolean isIntergenic = true;
         for (Gene currentGene : geneList) {

diff --git a/...rc/main/java/org/opencb/cellbase/core/variant/annotation/VariantAnnotationCalculator.java b/...rc/main/java/org/opencb/cellbase/core/variant/annotation/VariantAnnotationCalculator.java
@@ -442,7 +442,8 @@ private List<VariantAnnotation> runAnnotationProcess(List<Variant> normalizedVar
             if (annotatorSet.contains("consequenceType")) {
                 try {
                     List<ConsequenceType> consequenceTypeList = getConsequenceTypeList(normalizedVariantList.get(i),
-                        variantGeneList, true, QueryOptions.empty());
+                        variantGeneList, true, new QueryOptions().append("imprecise", imprecise)
+                                    .append("cnvExtraPadding", cnvExtraPadding));
                     variantAnnotation.setConsequenceTypes(consequenceTypeList);
                     if (phased) {
                         checkAndAdjustPhasedConsequenceTypes(normalizedVariantList.get(i), variantBuffer);

diff --git a/cellbase-lib/pom.xml b/cellbase-lib/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <groupId>org.opencb.cellbase</groupId>
         <artifactId>cellbase</artifactId>
-        <version>4.11.1</version>
+        <version>4.12.0-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

diff --git a/cellbase-lib/src/test/java/org/opencb/cellbase/lib/impl/VariantAnnotationCalculatorTest.java b/cellbase-lib/src/test/java/org/opencb/cellbase/lib/impl/VariantAnnotationCalculatorTest.java
@@ -1983,9 +1983,6 @@ public void testMissingClinVar() throws Exception {
         queryOptions.put("imprecise", true);
         queryOptions.put("phased", false);
 
-        QueryResult<VariantAnnotation> queryResult = variantAnnotationCalculator.getAnnotationByVariant(variant1, queryOptions);
-
-        assertEquals(14,new HashSet<String>(queryResult.getResult().get(0).getHgvs()).size());
 
     }
 
@@ -2020,6 +2017,39 @@ public void testClinicalVariantsGrch38() throws Exception {
 
     }
 
+    @Test
+    // long variant should have same annotations as short variant + cnvPadding
+    public void testCNVExtraPadding() throws Exception {
+
+        initGrch38();
+
+        // short variant
+        Variant shortVariant = new Variant("chr17:43087248-43087249:<CNV>");
+        // long variant
+        Variant longVariant = new Variant("chr17:43087248-43098506:<CNV>");
+
+        QueryOptions queryOptions = new QueryOptions("useCache", false);
+        queryOptions.put("include", "consequenceType, reference, alternate ,clinical");
+        queryOptions.put("normalize", true);
+        queryOptions.put("skipDecompose", false);
+        queryOptions.put("checkAminoAcidChange", true);
+        queryOptions.put("imprecise", true);
+        queryOptions.put("phased", false);
+
+        // long variant
+        QueryResult<VariantAnnotation> queryResult = variantAnnotationCalculator.getAnnotationByVariant(longVariant, queryOptions);
+        assertEquals(27, queryResult.getResult().get(0).getConsequenceTypes().size());
+
+        // short variant
+        queryResult = variantAnnotationCalculator.getAnnotationByVariant(shortVariant, queryOptions);
+        assertEquals(17, queryResult.getResult().get(0).getConsequenceTypes().size());
+
+        queryOptions.put("cnvExtraPadding", 10000);
+
+        // short variant with padding
+        queryResult = variantAnnotationCalculator.getAnnotationByVariant(shortVariant, queryOptions);
+        assertEquals(27, queryResult.getResult().get(0).getConsequenceTypes().size());
+    }
 
     // threw an exception
     @Test

diff --git a/cellbase-server/pom.xml b/cellbase-server/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <groupId>org.opencb.cellbase</groupId>
         <artifactId>cellbase</artifactId>
-        <version>4.11.1</version>
+        <version>4.12.0-SNAPSHOT</version>
         <relativePath>../pom.xml</relativePath>
     </parent>
 

diff --git a/cellbase-test/pom.xml b/cellbase-test/pom.xml
@@ -22,7 +22,7 @@
 
     <groupId>org.opencb.cellbase</groupId>
     <artifactId>cellbase-test</artifactId>
-    <version>4.11.1</version>
+    <version>4.12.0-SNAPSHOT</version>
     <packaging>pom</packaging>
 
     <dependencies>

diff --git a/pom.xml b/pom.xml
@@ -6,7 +6,7 @@
 
     <groupId>org.opencb.cellbase</groupId>
     <artifactId>cellbase</artifactId>
-    <version>4.11.1</version>
+    <version>4.12.0-SNAPSHOT</version>
     <packaging>pom</packaging>
 
     <name>CellBase project</name>
@@ -22,7 +22,7 @@
     </modules>
 
     <properties>
-        <cellbase.version>4.11.1</cellbase.version>
+        <cellbase.version>4.12.0-SNAPSHOT</cellbase.version>
         <compileSource>1.8</compileSource>
         <java-common-libs.version>3.7.5</java-common-libs.version>
         <biodata.version>1.5.6</biodata.version>