Merge branch 'develop' into feature/JAL-3187linkedFeatures

[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..1ac9ad7 100644 (file)
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -1,6 +1,5 @@
 package jalview.io.vcf;
 
-import jalview.analysis.AlignmentUtils;
 import jalview.analysis.Dna;
 import jalview.api.AlignViewControllerGuiI;
 import jalview.bin.Cache;
@@ -100,10 +99,12 @@ public class VCFLoader
    */
   private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
 
-  private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh38,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
+  private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
 
   private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
 
+  private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
+
   /*
    * keys to fields of VEP CSQ consequence data
    * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
@@ -262,8 +263,13 @@ public class VCFLoader
    */
   public SequenceI loadVCFContig(String contig)
   {
-    String ref = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY)
-            .getValue();
+    VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
+    if (headerLine == null)
+    {
+      Cache.log.error("VCF reference header not found");
+      return null;
+    }
+    String ref = headerLine.getValue();
     if (ref.startsWith("file://"))
     {
       ref = ref.substring(7);
@@ -282,7 +288,7 @@ public class VCFLoader
     }
     else
     {
-      System.err.println("VCF reference not found: " + ref);
+      Cache.log.error("VCF reference not found: " + ref);
     }
 
     return seq;
@@ -301,7 +307,7 @@ public class VCFLoader
     {
       VCFHeaderLine ref = header
               .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
-      String reference = ref.getValue();
+      String reference = ref == null ? null : ref.getValue();
 
       setSpeciesAndAssembly(reference);
 
@@ -373,7 +379,13 @@ public class VCFLoader
    */
   protected void setSpeciesAndAssembly(String reference)
   {
-    vcfSpecies = DEFAULT_SPECIES;
+    if (reference == null)
+    {
+      Cache.log.error("No VCF ##reference found, defaulting to "
+              + DEFAULT_REFERENCE + ":" + DEFAULT_SPECIES);
+      reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
+    }
+    reference = reference.toLowerCase();
 
     /*
      * for a non-human species, or other assembly identifier,
@@ -396,6 +408,7 @@ public class VCFLoader
       }
     }
 
+    vcfSpecies = DEFAULT_SPECIES;
     prop = Cache.getProperty(VCF_SPECIES);
     if (prop != null)
     {
@@ -640,7 +653,8 @@ public class VCFLoader
         /*
          * dna-to-peptide product mapping
          */
-        AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
+        // JAL-3187 render on the fly instead
+        // AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
       }
       else
       {
@@ -723,7 +737,7 @@ public class VCFLoader
     String species = seqCoords.getSpeciesId();
     String chromosome = seqCoords.getChromosomeId();
     String seqRef = seqCoords.getAssemblyId();
-    MapList map = seqCoords.getMap();
+    MapList map = seqCoords.getMapping();
 
     // note this requires the configured species to match that
     // returned with the Ensembl sequence; todo: support aliases?
@@ -803,40 +817,6 @@ 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.