From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Mon, 13 Nov 2017 16:57:02 +0000 (+0000)
Subject: JAL-2738 stubs for filtering VEP and VCF fields of interest
X-Git-Tag: Release_2_11_0~136
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=d8d9f50810cf347315bab5b7267416b3e5a8290d;p=jalview.git

JAL-2738 stubs for filtering VEP and VCF fields of interest
---

diff --git a/src/jalview/io/vcf/VCFLoader.java b/src/jalview/io/vcf/VCFLoader.java
index 5adc55c..e7c5a34 100644
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -61,16 +61,14 @@ public class VCFLoader
   /*
    * default VCF INFO key for VEP consequence data
    * NB this can be overridden running VEP with --vcf_info_field
-   * - we don't handle this case (require CSQ identifier)
+   * - we don't handle this case (require identifier to be CSQ)
    */
-  private static final String CSQ = "CSQ";
+  private static final String CSQ_FIELD = "CSQ";
 
   /*
-   * separator for fields in consequence data
+   * separator for fields in consequence data is '|'
    */
-  private static final String PIPE = "|";
-
-  private static final String PIPE_REGEX = "\\" + PIPE;
+  private static final String PIPE_REGEX = "\\|";
 
   /*
    * key for Allele Frequency output by VEP
@@ -126,6 +124,10 @@ public class VCFLoader
    */
   private String sourceId;
 
+  List<String> vcfFieldsOfInterest;
+
+  List<String> vepFieldsOfInterest;
+
   /**
    * Constructor given an alignment context
    * 
@@ -253,11 +255,13 @@ public class VCFLoader
    * Reads metadata (such as INFO field descriptions and datatypes) and saves
    * them for future reference
    * 
-   * @param sourceId
+   * @param theSourceId
    */
-  void saveMetadata(String sourceId)
+  void saveMetadata(String theSourceId)
   {
-    FeatureSource metadata = new FeatureSource(sourceId);
+    vcfFieldsOfInterest = new ArrayList<>();
+
+    FeatureSource metadata = new FeatureSource(theSourceId);
 
     for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
     {
@@ -285,9 +289,39 @@ public class VCFLoader
       }
       metadata.setAttributeName(attributeId, desc);
       metadata.setAttributeType(attributeId, attType);
+
+      if (isVcfFieldWanted(attributeId))
+      {
+        vcfFieldsOfInterest.add(attributeId);
+      }
     }
 
-    FeatureSources.getInstance().addSource(sourceId, metadata);
+    FeatureSources.getInstance().addSource(theSourceId, metadata);
+  }
+
+  /**
+   * Answers true if the VCF id is one we wish to capture in Jalview, else false
+   * 
+   * @param id
+   * @return
+   */
+  private boolean isVcfFieldWanted(String id)
+  {
+    // TODO option to match patterns in a Preferences entry?
+    return true;
+  }
+
+  /**
+   * Answers true if the VEP (CSQ) id is one we wish to capture in Jalview, else
+   * false
+   * 
+   * @param id
+   * @return
+   */
+  private boolean isVepFieldWanted(String id)
+  {
+    // TODO option to match patterns in a Preferences entry?
+    return true;
   }
 
   /**
@@ -298,7 +332,9 @@ public class VCFLoader
    */
   protected void locateCsqFields()
   {
-    VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ);
+    vepFieldsOfInterest = new ArrayList<>();
+
+    VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
     if (csqInfo == null)
     {
       return;
@@ -332,6 +368,12 @@ public class VCFLoader
         {
           csqFeatureFieldIndex = index;
         }
+
+        if (isVepFieldWanted(field))
+        {
+          vepFieldsOfInterest.add(field);
+        }
+
         index++;
       }
     }
@@ -702,13 +744,21 @@ public class VCFLoader
        * extract Consequence data (if present) that we are able to
        * associated with the allele for this variant feature
        */
-      if (CSQ.equals(key))
+      if (CSQ_FIELD.equals(key))
       {
         addConsequences(variant, seq, sf, altAlelleIndex);
         continue;
       }
 
       /*
+       * filter out fields we don't want to capture
+       */
+      if (!vcfFieldsOfInterest.contains(key))
+      {
+        continue;
+      }
+
+      /*
        * we extract values for other data which are allele-specific; 
        * these may be per alternate allele (INFO[key].Number = 'A') 
        * or per allele including reference (INFO[key].Number = 'R') 
@@ -775,7 +825,7 @@ public class VCFLoader
   protected void addConsequences(VariantContext variant, SequenceI seq,
           SequenceFeature sf, int altAlelleIndex)
   {
-    Object value = variant.getAttribute(CSQ);
+    Object value = variant.getAttribute(CSQ_FIELD);
 
     if (value == null || !(value instanceof ArrayList<?>))
     {
@@ -811,6 +861,8 @@ public class VCFLoader
     StringBuilder sb = new StringBuilder(128);
     boolean found = false;
 
+    // todo check against vepFieldsOfInterest as well somewhere
+
     for (String consequence : consequences)
     {
       String[] csqFields = consequence.split(PIPE_REGEX);
@@ -829,7 +881,7 @@ public class VCFLoader
 
     if (found)
     {
-      sf.setValue(CSQ, sb.toString());
+      sf.setValue(CSQ_FIELD, sb.toString());
     }
   }