From e8f7e6d965ef747995b02ee74ed3ba1e3a6cdb06 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Fri, 20 Nov 2020 11:59:09 +0000
Subject: [PATCH] JAL-3761 extra check that VCF feature is fully mappable to
 sequence

---
 src/jalview/io/vcf/VCFLoader.java |   53 ++++++++++++++++++++-----------------
 1 file changed, 29 insertions(+), 24 deletions(-)

diff --git a/src/jalview/io/vcf/VCFLoader.java b/src/jalview/io/vcf/VCFLoader.java
index f4ffc0c..f965c4e 100644
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -20,27 +20,6 @@
  */
 package jalview.io.vcf;
 
-import jalview.analysis.Dna;
-import jalview.api.AlignViewControllerGuiI;
-import jalview.bin.Cache;
-import jalview.datamodel.DBRefEntry;
-import jalview.datamodel.GeneLociI;
-import jalview.datamodel.Mapping;
-import jalview.datamodel.SequenceFeature;
-import jalview.datamodel.SequenceI;
-import jalview.datamodel.features.FeatureAttributeType;
-import jalview.datamodel.features.FeatureSource;
-import jalview.datamodel.features.FeatureSources;
-import jalview.ext.ensembl.EnsemblMap;
-import jalview.ext.htsjdk.HtsContigDb;
-import jalview.ext.htsjdk.VCFReader;
-import jalview.io.gff.Gff3Helper;
-import jalview.io.gff.SequenceOntologyI;
-import jalview.util.MapList;
-import jalview.util.MappingUtils;
-import jalview.util.MessageManager;
-import jalview.util.StringUtils;
-
 import java.io.File;
 import java.io.IOException;
 import java.util.ArrayList;
@@ -67,6 +46,26 @@ import htsjdk.variant.vcf.VCFHeaderLine;
 import htsjdk.variant.vcf.VCFHeaderLineCount;
 import htsjdk.variant.vcf.VCFHeaderLineType;
 import htsjdk.variant.vcf.VCFInfoHeaderLine;
+import jalview.analysis.Dna;
+import jalview.api.AlignViewControllerGuiI;
+import jalview.bin.Cache;
+import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.GeneLociI;
+import jalview.datamodel.Mapping;
+import jalview.datamodel.SequenceFeature;
+import jalview.datamodel.SequenceI;
+import jalview.datamodel.features.FeatureAttributeType;
+import jalview.datamodel.features.FeatureSource;
+import jalview.datamodel.features.FeatureSources;
+import jalview.ext.ensembl.EnsemblMap;
+import jalview.ext.htsjdk.HtsContigDb;
+import jalview.ext.htsjdk.VCFReader;
+import jalview.io.gff.Gff3Helper;
+import jalview.io.gff.SequenceOntologyI;
+import jalview.util.MapList;
+import jalview.util.MappingUtils;
+import jalview.util.MessageManager;
+import jalview.util.StringUtils;
 
 /**
  * A class to read VCF data (using the htsjdk) and add variants as sequence
@@ -674,7 +673,6 @@ public class VCFLoader
 
   /**
    * Transfers VCF features to sequences to which this sequence has a mapping.
-   * If the mapping is 3:1, computes peptide variants from nucleotide variants.
    * 
    * @param seq
    */
@@ -897,12 +895,19 @@ public class VCFLoader
           int[] featureRange = map.map.locateInFrom(variant.getStart(),
                   variant.getEnd());
 
+          /*
+           * only take features whose range is fully mappable to sequence positions
+           */
           if (featureRange != null)
           {
             int featureStart = Math.min(featureRange[0], featureRange[1]);
             int featureEnd = Math.max(featureRange[0], featureRange[1]);
-            count += addAlleleFeatures(seq, variant, featureStart,
-                    featureEnd, forwardStrand);
+            if (featureEnd - featureStart == variant.getEnd()
+                    - variant.getStart())
+            {
+              count += addAlleleFeatures(seq, variant, featureStart,
+                      featureEnd, forwardStrand);
+            }
           }
         }
         variants.close();
-- 
1.7.10.2