From: Mungo Carstairs <gmungoc@gmail.com>
Date: Sun, 27 Oct 2019 17:45:08 +0000 (+0000)
Subject: JAL-3469 add "Number=." fields (all values) to all alleles in VCF record
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=refs%2Fheads%2Ffeature%2FJAL-3469clinvarVCF;p=jalview.git

JAL-3469 add "Number=." fields (all values) to all alleles in VCF record
---

diff --git a/src/jalview/io/vcf/VCFLoader.java b/src/jalview/io/vcf/VCFLoader.java
index ac707d8..d3afc57 100644
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -19,6 +19,7 @@ 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;
@@ -247,9 +248,9 @@ public class VCFLoader
   private Set<String> badData;
 
   /**
-   * Constructor given a VCF file
+   * Constructor given a path to a VCF file
    * 
-   * @param alignment
+   * @param vcfFile
    */
   public VCFLoader(String vcfFile)
   {
@@ -912,7 +913,11 @@ public class VCFLoader
   }
 
   /**
-   * A convenience method to get an attribute value for an alternate allele
+   * A convenience method to get an attribute value for an alternate allele.
+   * {@code alleleIndex} is the position in the list of values for the allele.
+   * If {@alleleIndex == -1} then all values are concatenated (comma-separated).
+   * This is the case for fields declared with "Number=." i.e. values are not
+   * related to specific alleles.
    * 
    * @param variant
    * @param attributeName
@@ -924,16 +929,25 @@ public class VCFLoader
   {
     Object att = variant.getAttribute(attributeName);
 
+    String result = null;
     if (att instanceof String)
     {
-      return (String) att;
+      result = (String) att;
     }
-    else if (att instanceof ArrayList)
+    else if (att instanceof List<?>)
     {
-      return ((List<String>) att).get(alleleIndex);
+      List<String> theList = (List<String>) att;
+      if (alleleIndex == -1)
+      {
+        result = StringUtils.listToDelimitedString(theList, ",");
+      }
+      else
+      {
+        result = theList.get(alleleIndex);
+      }
     }
 
-    return null;
+    return result;
   }
 
   /**
@@ -1322,6 +1336,10 @@ public class VCFLoader
          */
         index++;
       }
+      else if (number == VCFHeaderLineCount.UNBOUNDED) // .
+      {
+        index = -1;
+      }
       else if (number != VCFHeaderLineCount.A)
       {
         /*
diff --git a/test/jalview/io/vcf/VCFLoaderTest.java b/test/jalview/io/vcf/VCFLoaderTest.java
index 87cf727..c050029 100644
--- a/test/jalview/io/vcf/VCFLoaderTest.java
+++ b/test/jalview/io/vcf/VCFLoaderTest.java
@@ -61,16 +61,17 @@ public class VCFLoaderTest
           + ">transcript4/1-18\n-----TGG-GGACGAGAGTGTGA-A\n";
 
   private static final String[] VCF = { "##fileformat=VCFv4.2",
-      // fields other than AF are ignored when parsing as they have no INFO definition
+      // note fields with no INFO definition are ignored when parsing
       "##INFO=<ID=AF,Number=A,Type=Float,Description=\"Allele Frequency, for each ALT allele, in the same order as listed\">",
       "##INFO=<ID=AC_Female,Number=A,Type=Integer,Description=\"Allele count in Female genotypes\"",
       "##INFO=<ID=AF_AFR,Number=A,Type=Float,Description=\"Allele Frequency among African/African American genotypes\"",
+      "##INFO=<ID=CLNSIG,Number=.,Type=String,Description=\"Clinical significance for this single variant\"",
       "##reference=Homo_sapiens/GRCh38",
       "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO",
       // A/T,C variants in position 2 of gene sequence (precedes transcript)
       // should create 2 variant features with respective AF values
       // malformed values for AC_Female and AF_AFR should be ignored
-      "17\t45051611\trs384765\tA\tT,C\t1666.64\tRF;XYZ\tAC=15;AF=5.0e-03,4.0e-03;AC_Female=12,3d;AF_AFR=low,2.3e-4",
+      "17\t45051611\trs384765\tA\tT,C\t1666.64\tRF;XYZ\tAC=15;AF=5.0e-03,4.0e-03;AC_Female=12,3d;AF_AFR=low,2.3e-4;CLNSIG=benign,probably_benign",
       // SNP G/C in position 4 of gene sequence, position 2 of transcript
       // insertion G/GA is transferred to nucleotide but not to peptide
       "17\t45051613\t.\tG\tGA,C\t1666.65\t.\tAC=15;AF=3.0e-03,2.0e-03",
@@ -133,6 +134,10 @@ public class VCFLoaderTest
     assertEquals(sf.getValue("ID"), "rs384765");
     assertEquals(sf.getValue("QUAL"), "1666.64");
     assertEquals(sf.getValue("FILTER"), "RF;XYZ");
+    /*
+     * if INFO declares Number=1, all values are attached to each allele
+     */
+    assertEquals(sf.getValue("CLNSIG"), "benign,probably_benign");
     // malformed integer for AC_Female is ignored (JAL-3375)
     assertNull(sf.getValue("AC_Female"));
 
@@ -148,6 +153,7 @@ public class VCFLoaderTest
     // malformed float for AF_AFR is ignored (JAL-3375)
     assertNull(sf.getValue("AC_AFR"));
     assertEquals(sf.getValue(Gff3Helper.ALLELES), "A,T");
+    assertEquals(sf.getValue("CLNSIG"), "benign,probably_benign");
 
     sf = geneFeatures.get(2);
     assertEquals(sf.getFeatureGroup(), "VCF");