Handle multiple input VCFs in annotate-svs (#75)

Closes: #75
Related-Issue: #75
Projected-Results-Impact: none

CHANGES.md

@@ -5,6 +5,7 @@
 - Writing out proper SV type for Dragen CNV (#76)
 - Adding support for depth of coverage annotation (#73)
 - Ensure output files are sorted by chromosomes (#79)
+- Merge multiple input VCFs in annotate-svs with clustering (#75)
 
 ## v0.26
 

pom.xml

@@ -25,6 +25,7 @@
 		<htsjdk.version>2.24.1</htsjdk.version>
 		<jannovar.version>0.41</jannovar.version>
 		<externalsortinginjava.version>0.6.1</externalsortinginjava.version>
+		<json.version>20220320</json.version>
 	</properties>
 
 	<modules>

tests/hg19-chr22/Case_1_index.delly2.gts.tsv-expected

@@ -1,2 +1,2 @@
 release	chromosome	chromosome_no	bin	chromosome2	chromosome_no2	bin2	pe_orientation	start	end	start_ci_left	start_ci_right	end_ci_left	end_ci_right	case_id	set_id	sv_uuid	caller	sv_type	sv_sub_type	info	num_hom_alt	num_hom_ref	num_het	num_hemi_alt	num_hemi_ref	genotype
-GRCh37	22	22	89	22	22	89	3to5	17400000	17700000	-29	29	-29	29	.	.	UUID	EMBL.DELLYv1.1.3	DEL	DEL	{"""affectedCarriers""":0,"""backgroundCarriers""":0,"""unaffectedCarriers""":0}	0	2	1	0	0	{"""Case_1_father-N1-DNA1-WGS1""":{"""cn""":2,"""ft""":{"""LowQual"""},"""gq""":14,"""gt""":"""0/1""","""pec""":0,"""pev""":0,"""src""":34,"""srv""":4},"""Case_1_index-N1-DNA1-WGS1""":{"""cn""":2,"""gq""":35,"""gt""":"""0/0""","""pec""":0,"""pev""":0,"""src""":29,"""srv""":2},"""Case_1_mother-N1-DNA1-WGS1""":{"""cn""":2,"""gq""":67,"""gt""":"""0/0""","""pec""":0,"""pev""":0,"""src""":32,"""srv""":1}}
+GRCh37	22	22	89	22	22	89	3to5	17400000	17700000	-29	29	-29	29	.	.	UUID	EMBL.DELLYv1.1.3	DEL	DEL	{"""affectedCarriers""":0,"""backgroundCarriers""":0,"""unaffectedCarriers""":0}	0	2	1	0	0	{"""Case_1_father-N1-DNA1-WGS1""":{"""cn""":2,"""ft""":["""LowQual"""],"""gq""":14,"""gt""":"""0/1""","""pec""":0,"""pev""":0,"""src""":34,"""srv""":4},"""Case_1_index-N1-DNA1-WGS1""":{"""cn""":2,"""gq""":35,"""gt""":"""0/0""","""pec""":0,"""pev""":0,"""src""":29,"""srv""":2},"""Case_1_mother-N1-DNA1-WGS1""":{"""cn""":2,"""gq""":67,"""gt""":"""0/0""","""pec""":0,"""pev""":0,"""src""":32,"""srv""":1}}

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/annotate_svs/AnnotateSvsArgs.java

@@ -57,9 +57,12 @@ public final class AnnotateSvsArgs {
 
   @Parameter(
       names = "--input-vcf",
-      description = "Path to input VCF file to annotate",
+      description =
+          "Path to input VCF file to annotate.  You can specify this multiple times to read in files from "
+              + "different callers, for example.  In this case, the variants will be merged based on the "
+              + "--merge-* arguments.",
       required = true)
-  private String inputVcf;
+  private List<String> inputVcf = new ArrayList<>();
 
   @Parameter(
       names = "--input-ped",
@@ -118,6 +121,16 @@ public final class AnnotateSvsArgs {
           "Annotate CNV with coverage and mapping quality from maelstrom-core coverage VCF file")
   private List<String> coverageVcfs = new ArrayList<>();
 
+  @Parameter(
+      names = "--merge-overlap",
+      description = "Reciprocal overlap to require for merging (default: 0.75)")
+  private double mergeOverlap = 0.75;
+
+  @Parameter(
+      names = "--merge-bnd-radius",
+      description = "Merge BNDs within the given radius (default: 50)")
+  private int mergeBndRadius = 50;
+
   public String getRefseqSerPath() {
     return refseqSerPath;
   }
@@ -150,7 +163,7 @@ public String getSetId() {
     return setId;
   }
 
-  public String getInputVcf() {
+  public List<String> getInputVcf() {
     return inputVcf;
   }
 
@@ -198,6 +211,14 @@ public List<String> getCoverageVcfs() {
     return coverageVcfs;
   }
 
+  public double getMergeOverlap() {
+    return mergeOverlap;
+  }
+
+  public int getMergeBndRadius() {
+    return mergeBndRadius;
+  }
+
   @Override
   public String toString() {
     return "AnnotateSvsArgs{"
@@ -259,6 +280,10 @@ public String toString() {
         + '\''
         + ", coverageVcfs="
         + coverageVcfs
+        + ", mergeOverlap="
+        + mergeOverlap
+        + ", mergeBndRadius="
+        + mergeBndRadius
         + '}';
   }
 }