Allow rename of SnpSift fields in MultiSourceAnnotator (#287)

* Allow renaming of fields in MultiSourceAnnotator

src/main/java/com/github/discvrseq/walkers/ClinvarAnnotator.java

@@ -23,7 +23,7 @@
 /**
  * This tool compares an input VCF to the provided <a href=https://www.ncbi.nlm.nih.gov/clinvar/>ClinVar</a> VCF and adds annotations to
  * any variant overlapping with a variant from ClinVar (matching both site and allele). Note, this requires the VCF to use ClinVar's 2.0 VCF format.
- * The ClinVar VCFs can be foind here: <a href="https://www.ncbi.nlm.nih.gov/variation/docs/ClinVar_vcf_files/">https://www.ncbi.nlm.nih.gov/variation/docs/ClinVar_vcf_files/</a>
+ * The ClinVar VCFs can be found here: <a href="https://www.ncbi.nlm.nih.gov/variation/docs/ClinVar_vcf_files/">https://www.ncbi.nlm.nih.gov/variation/docs/ClinVar_vcf_files/</a>
  * <p/>
  *
  * <h3>Usage example</h3>
@@ -60,7 +60,7 @@ public class ClinvarAnnotator extends VariantWalker {
     /**
      * INFO header fields in the new VCF for ClinVar annotations.
      */
-    private List<VCFInfoHeaderLine> HEADER_LINES = Arrays.asList(
+    private final List<VCFInfoHeaderLine> HEADER_LINES = Arrays.asList(
             new VCFInfoHeaderLine("CLN_ALLELE", VCFHeaderLineCount.R, VCFHeaderLineType.Character, "Alternate alleles from Clinvar"),
             new VCFInfoHeaderLine("CLN_ALLELEID", VCFHeaderLineCount.R, VCFHeaderLineType.Integer, "the ClinVar Allele ID"),
             new VCFInfoHeaderLine("CLN_DN", VCFHeaderLineCount.R, VCFHeaderLineType.String, "ClinVar's preferred disease name for the concept specified by disease identifiers in CLNDISDB"),
@@ -120,7 +120,7 @@ public void apply(VariantContext variant, ReadsContext readsContext, ReferenceCo
                 if (cvAllele.equals(alt)){
                     //gather annotations, add to map by allele
                     foundHit = true;
-                    annotationMap.put(alt, transferAnnotations(vc, alt, vcb));
+                    annotationMap.put(alt, transferAnnotations(vc, alt));
                 }
             }
         }
@@ -148,7 +148,7 @@ public void apply(VariantContext variant, ReadsContext readsContext, ReferenceCo
             if (nonNull > 0){
                 if (!sb.isEmpty()){
                     hasAnnotation = true;
-                    vcb.attribute(line.getID(), sb.stream().collect(Collectors.joining(",")));
+                    vcb.attribute(line.getID(), String.join(",", sb));
                 }
             }
         }
@@ -187,11 +187,10 @@ public void apply(VariantContext variant, ReadsContext readsContext, ReferenceCo
      * </p>
      *
      * @param source    ClinVar VCF
-     * @param alt   alternate allele from ClinVar
-     * @param vcb   new VCF to annotate
+     * @param alt   alternate allele from ClinVar*
      * @return  a map which maps ClinVar alleles with their corresponding ClinVar annotations
      */
-    private Map<String, String> transferAnnotations (VariantContext source, Allele alt, VariantContextBuilder vcb){
+    private Map<String, String> transferAnnotations(VariantContext source, Allele alt){
         Map<String, String> annotations = new HashMap<>();
 
         annotations.put("CLN_ALLELE", alt.getDisplayString());
@@ -229,7 +228,6 @@ private Map<String, String> transferAnnotations (VariantContext source, Allele a
      *
      * @param source    ClinVar VCF
      * @param ID   ClinVar VCF Annotation ID
-     * @return
      */
     private String annotateValue(VariantContext source, String ID){
         if (source.getAttribute(ID) == null){

src/main/java/com/github/discvrseq/walkers/MultiSourceAnnotator.java

@@ -18,9 +18,11 @@
 import org.broadinstitute.hellbender.exceptions.GATKException;
 import org.broadinstitute.hellbender.utils.Utils;
 
+import javax.annotation.Nullable;
 import java.io.File;
 import java.util.*;
 import java.util.stream.Collectors;
+import java.util.stream.IntStream;
 
 /**
  * This is a fairly specialized tool, designed to take the VCFs annotated with ClinvarAnnotator (from DISCVR-seq Toolkit) and <a href="https://www.hgsc.bcm.edu/software/cassandra">Cassandra</a>,
@@ -66,6 +68,11 @@ public class MultiSourceAnnotator extends VariantWalker {
     @Argument(doc="SnpSift Allowed INFO Fields", fullName = "snpsift-fields", shortName = "ssf", optional = true)
     public List<String> snpSiftFields = new ArrayList<>(SNPSIFT_INFO);
 
+    @Argument(doc="Renamed SnpSift INFO Fields. If provided, this must be of equal length as snpsift-fields. Each source snpsift-field will be transferred to an annotation of this name", fullName = "renamed-snpsift-fields", shortName = "rssf", optional = true)
+    public List<String> renamedSnpSiftFields = new ArrayList<>();
+
+    final Map<String, String> snpSiftFieldMapping = new HashMap<>();
+
     @Argument(doc="Funcotator Annotated VCF", fullName = "funcotator", shortName = "f", optional = true)
     public FeatureInput<VariantContext> funcotatorVariants = null;
 
@@ -466,6 +473,14 @@ public void onTraversalStart() {
                 allAnnotationKeys.add(id);
             }
 
+            if (renamedSnpSiftFields != null && !renamedSnpSiftFields.isEmpty()) {
+                if (renamedSnpSiftFields.size() != snpSiftFields.size()) {
+                    throw new GATKException("--renamed-snpsift-fields must be the same length as --snpsift-fields");
+                }
+
+                IntStream.range(0, renamedSnpSiftFields.size()).forEach(j -> snpSiftFieldMapping.put(snpSiftFields.get(j), renamedSnpSiftFields.get(j)));
+            }
+
             List<String> allKeys = new ArrayList<>(snpSiftHeader.getInfoHeaderLines().stream().map(VCFInfoHeaderLine::getID).toList());
             allKeys.removeAll(snpSiftFields);
             if (!allKeys.isEmpty()) {
@@ -533,7 +548,7 @@ public void apply(VariantContext variant, ReadsContext readsContext, ReferenceCo
                 }
 
                 snpSift++;
-                transferInfoData(vcb, vc, snpSiftFields);
+                transferInfoData(vcb, vc, snpSiftFields, snpSiftFieldMapping);
             }
         }
 
@@ -593,9 +608,18 @@ private boolean matches(VariantContext source, VariantContext annotation){
     }
 
     private void transferInfoData(VariantContextBuilder vcb, VariantContext source, List<String> ids){
+        transferInfoData(vcb, source, ids, null);
+    }
+
+    private void transferInfoData(VariantContextBuilder vcb, VariantContext source, List<String> ids, @Nullable Map<String, String> oldKeyToNewKey){
         for (String id : ids){
             if (source.hasAttribute(id) && source.getAttribute(id) != null){
-                vcb.attribute(id, source.getAttribute(id));
+                if (oldKeyToNewKey != null) {
+                    vcb.attribute(oldKeyToNewKey.getOrDefault(id, id), source.getAttribute(id));
+                }
+                else {
+                    vcb.attribute(id, source.getAttribute(id));
+                }
             }
         }
     }