SVAnnotate: Functional annotations for SVs called by GATK-SV (#7431)

broadinstitute · Mar 9, 2022 · 1c749b3 · 1c749b3
1 parent c1190ac
commit 1c749b3
Show file tree

Hide file tree

Showing 19 changed files with 2,200 additions and 6 deletions.
diff --git a/src/main/java/org/broadinstitute/hellbender/tools/spark/sv/utils/GATKSVVCFConstants.java b/src/main/java/org/broadinstitute/hellbender/tools/spark/sv/utils/GATKSVVCFConstants.java
@@ -73,6 +73,8 @@ public final class GATKSVVCFConstants {
     public static final String CPX_EVENT_ALT_ARRANGEMENTS = "ALT_ARRANGEMENT";
     public static final String CPX_SV_REF_SEGMENTS = "SEGMENTS";
     public static final String CPX_EVENT_KEY = "CPX_EVENT";
+    public static final String CPX_INTERVALS = "CPX_INTERVALS";
+    public static final String CPX_TYPE = "CPX_TYPE";
 
     // not defined in output vcf header but used in internal id that is currently output in the ID column
     public static final String INTERVAL_VARIANT_ID_FIELD_SEPARATOR = "_";
@@ -86,6 +88,8 @@ public final class GATKSVVCFConstants {
     public static final String DEPTH_ALGORITHM = "depth";
     public static final String CONTIG2_ATTRIBUTE = "CHR2";
     public static final String END2_ATTRIBUTE = "END2";
+    public static final String BND_DELETION_STRANDS = "+-";
+    public static final String BND_DUPLICATION_STRANDS = "-+";
 
     // format block
     public static final String COPY_NUMBER_FORMAT = "CN";
@@ -101,6 +105,23 @@ public final class GATKSVVCFConstants {
     // Clustering
     public static final String CLUSTER_MEMBER_IDS_KEY = "MEMBERS";
 
+    // functional annotations
+    public static final String LOF = "PREDICTED_LOF";
+    public static final String INT_EXON_DUP = "PREDICTED_INTRAGENIC_EXON_DUP";
+    public static final String COPY_GAIN = "PREDICTED_COPY_GAIN";
+    public static final String DUP_PARTIAL = "PREDICTED_DUP_PARTIAL";
+    public static final String PARTIAL_EXON_DUP = "PREDICTED_PARTIAL_EXON_DUP";
+    public static final String INTRONIC = "PREDICTED_INTRONIC";
+    public static final String INV_SPAN = "PREDICTED_INV_SPAN";
+    public static final String UTR = "PREDICTED_UTR";
+    public static final String MSV_EXON_OVERLAP = "PREDICTED_MSV_EXON_OVERLAP";
+    public static final String PROMOTER = "PREDICTED_PROMOTER";
+    public static final String BREAKEND_EXON = "PREDICTED_BREAKEND_EXONIC";
+    public static final String INTERGENIC = "PREDICTED_INTERGENIC";
+    public static final String NONCODING_SPAN = "PREDICTED_NONCODING_SPAN";
+    public static final String NONCODING_BREAKPOINT = "PREDICTED_NONCODING_BREAKPOINT";
+    public static final String NEAREST_TSS = "PREDICTED_NEAREST_TSS";
+    public static final String TSS_DUP = "PREDICTED_TSS_DUP";
 
     ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////
 

diff --git a/src/main/java/org/broadinstitute/hellbender/tools/spark/sv/utils/SVUtils.java b/src/main/java/org/broadinstitute/hellbender/tools/spark/sv/utils/SVUtils.java
@@ -1,11 +1,13 @@
 package org.broadinstitute.hellbender.tools.spark.sv.utils;
 
 import htsjdk.samtools.*;
+import htsjdk.samtools.util.Locatable;
 import htsjdk.variant.variantcontext.VariantContext;
 import htsjdk.variant.vcf.VCFConstants;
 import org.broadinstitute.hellbender.exceptions.UserException;
 import org.broadinstitute.hellbender.tools.spark.utils.HopscotchSetSpark;
 import org.broadinstitute.hellbender.tools.spark.utils.LongIterator;
+import org.broadinstitute.hellbender.utils.SVInterval;
 import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.Utils;
 import org.broadinstitute.hellbender.utils.io.IOUtils;
@@ -109,6 +111,34 @@ public static Set<String> getCanonicalChromosomes(final String nonCanonicalConti
         }
     }
 
+    /**
+     * Get numeric contig ID from string contig name based on SAMSequenceDictionary
+     * @throws IllegalArgumentException if contig name is not in sequence dictionary
+     * @param contigName - string version of contig name
+     * @param sequenceDictionary - SAMSequenceDictionary to map from name to ID
+     * @return - Integer contig ID
+     */
+    public static Integer getContigIDFromName(final String contigName, final SAMSequenceDictionary sequenceDictionary) {
+        final Integer contigID = sequenceDictionary.getSequenceIndex(contigName);
+        if (contigID == -1) {
+            throw new IllegalArgumentException("Contig " + contigName + " not in provided contig ID to name map");
+        }
+        return contigID;
+    }
+
+
+    /**
+     * Converts locatable (uses 1-based, closed intervals) to SVInterval (1-based half-open). Conversion: end + 1
+     * @param loc - Locatable object (uses 1-based, closed intervals) to convert
+     * @param sequenceDictionary - SAMSequenceDictionary to map to numeric contig ID
+     * @return - SVInterval representing the same interval as the locatable, converted to 1-based half-open
+     */
+    public static SVInterval locatableToSVInterval(final Locatable loc,
+                                                   final SAMSequenceDictionary sequenceDictionary) {
+        final Integer contigID = getContigIDFromName(loc.getContig(), sequenceDictionary);
+        return new SVInterval(contigID, loc.getStart(), loc.getEnd() + 1);
+    }
+
     // =================================================================================================================
 
     /** return a good initialCapacity for a HashMap that will hold a given number of elements */