JAL-2738 JAL-2743 update unit test to match new VCF properties

[jalview.git] / src / jalview / io / vcf / VCFLoader.java

diff --git a/src/jalview/io/vcf/VCFLoader.java b/src/jalview/io/vcf/VCFLoader.java
index f4ce1a3..9831af7 100644 (file)
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -262,8 +262,8 @@ public class VCFLoader
    */
   public SequenceI loadVCFContig(String contig)
   {
-    String ref = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY)
-            .getValue();
+    VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
+    String ref = headerLine == null ? null : headerLine.getValue();
     if (ref.startsWith("file://"))
     {
       ref = ref.substring(7);
@@ -373,6 +373,7 @@ public class VCFLoader
    */
   protected void setSpeciesAndAssembly(String reference)
   {
+    reference = reference.toLowerCase();
     vcfSpecies = DEFAULT_SPECIES;
 
     /*
@@ -803,40 +804,6 @@ public class VCFLoader
   }
 
   /**
-   * Answers true if the species inferred from the VCF reference identifier
-   * matches that for the sequence
-   * 
-   * @param vcfAssembly
-   * @param speciesId
-   * @return
-   */
-  boolean vcfSpeciesMatchesSequence(String vcfAssembly, String speciesId)
-  {
-    // PROBLEM 1
-    // there are many aliases for species - how to equate one with another?
-    // PROBLEM 2
-    // VCF ##reference header is an unstructured URI - how to extract species?
-    // perhaps check if ref includes any (Ensembl) alias of speciesId??
-    // TODO ask the user to confirm this??
-
-    if (vcfAssembly.contains("Homo_sapiens") // gnomAD exome data example
-            && "HOMO_SAPIENS".equals(speciesId)) // Ensembl species id
-    {
-      return true;
-    }
-
-    if (vcfAssembly.contains("c_elegans") // VEP VCF response example
-            && "CAENORHABDITIS_ELEGANS".equals(speciesId)) // Ensembl
-    {
-      return true;
-    }
-
-    // this is not a sustainable solution...
-
-    return false;
-  }
-
-  /**
    * Queries the VCF reader for any variants that overlap the mapped chromosome
    * ranges of the sequence, and adds as variant features. Returns the number of
    * overlapping variants found.