Merge pull request #308 from biowdl/gridss

Adjust gridss SV typing task to also filter duplicate records

CHANGELOG.md

@@ -9,6 +9,15 @@ that users understand how the changes affect the new version.
 -->
 version 5.1.0-dev
 ---------------------------
++ Mateclever's runtime attribute defaults were changed to:
+  + memory: `"250GiB"`
+  + timeMinutes: `2880`
++ Clever's Prediction task's runtime attribute defaults were changed to:
+  + memory: `"80GiB"`
+  + timeMinutes: `2200`
++ The GRIDSS AnnotateSvTypes task now also removes the second breakend of
+  the breakpoints and single breakends. This will prepare the output better
+  to be passed into survivor.
 + Updated SURVIVOR version to 1.0.7
 + Add a combined samtools dict and samtools faidx task.
 + Add a BWA index task.

clever.wdl

@@ -34,8 +34,8 @@ task Mateclever {
         Int maxOffset = 150
 
         Int threads = 10
-        String memory = "15GiB"
-        Int timeMinutes = 600
+        String memory = "250GiB"
+        Int timeMinutes = 2880
         String dockerImage = "quay.io/biocontainers/clever-toolkit:2.4--py36hcfe0e84_6"
     }
 
@@ -94,8 +94,8 @@ task Prediction {
         String outputPath = "./clever"
 
         Int threads = 10
-        String memory = "55GiB"
-        Int timeMinutes = 480
+        String memory = "80GiB"
+        Int timeMinutes = 2200
         String dockerImage = "quay.io/biocontainers/clever-toolkit:2.4--py36hcfe0e84_6"
     }
 

gridss.wdl

@@ -119,9 +119,14 @@ task AnnotateSvTypes {
         gr <- breakpointRanges(vcf)
         svtype <- simpleEventType(gr)
         info(vcf[gr$sourceId])$SVTYPE <- svtype
-        # GRIDSS doesn't supply a GT, so we estimate GT based on AF (assuming CN of 2, might be inaccurate)
-        geno(vcf)$GT <- ifelse(geno(vcf)$AF > 0.75, "1/1", ifelse(geno(vcf)$AF < 0.25, "0/0", "0/1"))
-        writeVcf(vcf, out_path, index=~{index})
+        # GRIDSS doesn't supply a GT, simply set it to 0/1
+        geno(vcf)$GT <- as.matrix(sapply(row.names(vcf), function(x) {"0/1"}))
+        # Select only one breakend per event (also removes single breakends):
+        # sourceId ends with o or h for paired breakends, the first in the pair
+        # end with o the second with h. Single breakend end with b, these will
+        # also be removed since we can't determine the SVTYPE.
+        gr2 <- gr[grepl(".*o$", gr$sourceId)]
+        writeVcf(vcf[gr2$sourceId], out_path, index=~{index})
         EOF
     >>>