From: Sarah Wait Zaranek <swz@curii.com>
Date: Mon, 27 Jul 2020 19:56:49 +0000 (+0000)
Subject: Updating to allow for two vcfs of known sites
X-Git-Url: https://git.arvados.org/arvados-tutorial.git/commitdiff_plain/814e2840b16acbe0e52e6fb3095a87b33269dc50?hp=ec300c76b5a3f927c3cd5035669b70847ade3854

Updating to allow for two vcfs of known sites
Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek <swz@curii.com>
no issue #
---

diff --git a/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl b/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl
index dbb1d50..76e4798 100644
--- a/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl
+++ b/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl
@@ -32,12 +32,18 @@ inputs:
   sample: 
     type: string
     label: Sample Name
-  knownsites:
+  knownsites1:
     type: File
     format: edam:format_3016 # VCF
     label: VCF of known polymorphic sites for BQSR
     secondaryFiles:
-      - .tbi   
+      - .idx 
+  knownsites2:
+    type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
+    secondaryFiles:
+      - .tbi  
   scattercount: 
     type: string
     label: Desired split for variant calling
@@ -117,7 +123,8 @@ steps:
       bam: samtools-index/indexedbam
       sample: sample
       scattercount: scattercount
-      knownsites: knownsites
+      knownsites1: knownsites1
+      knownsites2: knownsites2
     out: [gatheredgvcf]
   generate-report:
     run: report-wf.cwl
diff --git a/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl b/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
index 7ae0837..2de30a6 100644
--- a/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
+++ b/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
@@ -41,7 +41,13 @@ inputs:
   sample:
     type: string
     label: Sample Name
-  knownsites:
+  knownsites1:
+    type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
+    secondaryFiles:
+      - .idx
+  knownsites2:
     type: File
     format: edam:format_3016 # VCF
     label: VCF of known polymorphic sites for BQSR
@@ -69,7 +75,9 @@ arguments:
   - prefix: "-I"
     valueFrom: $(inputs.bam)
   - prefix: "--known-sites"
-    valueFrom: $(inputs.knownsites)
+    valueFrom: $(inputs.knownsites1)
+  - prefix: "--known-sites"
+    valueFrom: $(inputs.knownsites2)
   - prefix: "-L"
     valueFrom: $(inputs.intervallist)
   - prefix: "-O"
diff --git a/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl b/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
index de4a648..d9023c9 100644
--- a/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
+++ b/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
@@ -27,7 +27,13 @@ inputs:
   sample:
     type: string
     label: Sample Name
-  knownsites:
+  knownsites1:
+    type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
+    secondaryFiles:
+      - .idx
+  knownsites2:
     type: File
     format: edam:format_3016 # VCF
     label: VCF of known polymorphic sites for BQSR
@@ -51,7 +57,8 @@ steps:
       bam: bam
       reference: reference
       sample: sample
-      knownsites: knownsites
+      knownsites1: knownsites1
+      knownsites2: knownsites2
       intervallist: intervallist
     out: [recaltable]
   applyBQSR:
diff --git a/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl b/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl
index 80d7601..13bd4af 100644
--- a/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl
+++ b/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl
@@ -31,7 +31,13 @@ inputs:
   sample:
     type: string
     label: Sample Name
-  knownsites:
+  knownsites1:
+    type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
+    secondaryFiles:
+      - .idx
+  knownsites2:
     type: File
     format: edam:format_3016 # VCF
     label: VCF of known polymorphic sites for BQSR
@@ -67,7 +73,8 @@ steps:
       bam: bam
       reference: reference
       sample: sample
-      knownsites: knownsites
+      knownsites1: knownsites1
+      knownsites2: knownsites2
       intervallist: splitintervals/intervalfiles
     out: [gvcf]
 
diff --git a/WGS-processing/cwl/wgs-processing-wf.cwl b/WGS-processing/cwl/wgs-processing-wf.cwl
index 8cbcf5e..8499b92 100644
--- a/WGS-processing/cwl/wgs-processing-wf.cwl
+++ b/WGS-processing/cwl/wgs-processing-wf.cwl
@@ -24,12 +24,18 @@ inputs:
       - ^.dict
   fullintervallist:
     type: File
-  knownsites:
+  knownsites1:
     type: File
     format: edam:format_3016 # VCF
-    label: VCF of known polymorphic sites for BQSR
+    label: VCF of known SNPS sites for BQSR
     secondaryFiles:
-      - .tbi   
+      - .idx   
+  knownsites2:
+    type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known indel sites for BQSR
+    secondaryFiles:
+      - .tbi
   scattercount: 
     type: string
     label: Desired split for variant calling
@@ -78,7 +84,8 @@ steps:
       reference: reference
       fullintervallist: fullintervallist
       sample: getfastq/sample
-      knownsites: knownsites
+      knownsites1: knownsites1
+      knownsites2: knownsites2
       scattercount: scattercount
       clinvarvcf: clinvarvcf
       reportfunc: reportfunc
diff --git a/WGS-processing/yml/wgs-processing-wf.yml b/WGS-processing/yml/wgs-processing-wf.yml
index 299a674..49ac46d 100644
--- a/WGS-processing/yml/wgs-processing-wf.yml
+++ b/WGS-processing/yml/wgs-processing-wf.yml
@@ -32,7 +32,12 @@ tailhtml:
 
 scattercount: '20'
 
-knownsites:
+knownsites1:
+  class: File
+  format: edam:format_3016
+  location: keep:7c0b13bda857fa15d88c1039182f69d5+8052/Homo_sapiens_assembly38.dbsnp138.vcf 
+
+knownsites2:
   class: File
   format: edam:format_3016
   location: keep:7c0b13bda857fa15d88c1039182f69d5+8052/Homo_sapiens_assembly38.known_indels.vcf.gz