From f8190b9075da8df466cd1faa59840619e9233b7c Mon Sep 17 00:00:00 2001
From: Sarah Wait Zaranek <swz@curii.com>
Date: Wed, 17 Jun 2020 17:13:54 +0000
Subject: [PATCH] Add scatter wf for overall scatter
 Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek <swz@curii.com> no issue #

---
 WGS-processing/cwl/getfastq.cwl               |  49 ------------
 .../.scatter-gatk-wf-with-interval.cwl.swp    | Bin 0 -> 12288 bytes
 ...eport-wf.cwl => bwamem-gatk-report-wf.cwl} |   0
 .../cwl/helper/{ => not-in-use}/getgvcfs.cwl  |   0
 WGS-processing/cwl/wgs-processing-wf.cwl      |  72 +++++-------------
 5 files changed, 20 insertions(+), 101 deletions(-)
 delete mode 100644 WGS-processing/cwl/getfastq.cwl
 create mode 100644 WGS-processing/cwl/helper/.scatter-gatk-wf-with-interval.cwl.swp
 rename WGS-processing/cwl/helper/{bwa-gatk-report-wf.cwl => bwamem-gatk-report-wf.cwl} (100%)
 rename WGS-processing/cwl/helper/{ => not-in-use}/getgvcfs.cwl (100%)

diff --git a/WGS-processing/cwl/getfastq.cwl b/WGS-processing/cwl/getfastq.cwl
deleted file mode 100644
index a38bc91..0000000
--- a/WGS-processing/cwl/getfastq.cwl
+++ /dev/null
@@ -1,49 +0,0 @@
-$namespaces:
-  arv: "http://arvados.org/cwl#"
-  cwltool: "http://commonwl.org/cwltool#"
-class: ExpressionTool
-cwlVersion: v1.1
-label: Create array of gvcfs to process
-requirements:
-  InlineJavascriptRequirement: {}
-inputs:
-  fastjdir:
-    type: Directory
-    label: Input directory of fastj
-    loadListing: 'shallow_listing' 
-outputs:
-  fastq1: 
-    type: File[]
-  fastq2:
-    type: File[]
-expression: |
-  ${function compare(a, b) {
-    var baseA = a.basename;
-    var baseB = b.basename;
-
-    var comparison = 0;
-    if (baseA > baseB) {
-    comparison = 1;
-    } else if (baseA < baseB) {
-    comparison = -1;
-    }
-    return comparison;
-    }
-
-    var fastq1 = [];
-    var fastq2 = [];
-    for (var i = 0; i < inputs.fastjdir.listing.length; i++) {
-      var name = inputs.fastjdir.listing[i];
-      if (name.basename.indexOf('_1.fastq.gz') != -1 ) {
-        fastq1.push(name);
-      }
-      if (name.basename.indexOf('_2.fastq.gz') != -1 ) {
-        fastq2.push(name);
-      }
-    }
-  
-    fastq1 = fastq1.sort(compare)
-    fastq2 = fastq2.sort(compare)
- 
-    return {"fastq1": fastq1, "fastq2": fastq2};
-  }
diff --git a/WGS-processing/cwl/helper/.scatter-gatk-wf-with-interval.cwl.swp b/WGS-processing/cwl/helper/.scatter-gatk-wf-with-interval.cwl.swp
new file mode 100644
index 0000000000000000000000000000000000000000..6973e8cc94c9851ec8dc8dbb4908a04331c0d92c
GIT binary patch
literal 12288
zcmeI2&5ImG7>8@UX+9La=^>Sk7nSbLMkS;V5)#<pAqcK+4#9<*?wXl$tGhZ?)jO^P
z#7n$*6VZ#}(Szby{RjMjC|)IhfB_F8n2QI0Ro!jRgf$DfN#2HMdaCNJ>Uy89g<bZY
zoZf!<GC!NXz;Hdr*iYwg?|k=Z{>R5Jt}!R9n1}vDF0E#BU#ohSlyYw)b#tk8>e_Q>
zbSmtgD2)@1N(+{i!i&U}W!B7^tRG5z>y1@HXICz5C5<(ObPj7}#X@ITsgX8wh45b5
zWG4J>vX~|d<>!g2(X}UZioioQNW)}+48XuK_WYTR(?P|l^^^RG$6qHaWPl8i0Wv@a
z$N(8217v^<JU|9~v&P;*z7O>|U+C{^cisCY-N*nLAOmE843GgbKnBPF86X2>fDDj<
z|Dgd<GWPewjPWDbJpTVb{Qdvuqm2CmJ_jFx_rMMC7I+0b3!VWd!IR(!xO1Gb+u#<s
z3BCbeg3rJQ;0AabRA2{O1}}jOJOz$}JI5IN3;Yglft%oK@CA4mTn8;Ezy*+l6W|f>
z+fl}T2H%6P!27_0H^EhK9y|w*f?tvEkKhOJE%*d{2o!h?oCCqh2|yPaAOmE843Ggb
zKnBRbgJXc>GOf?~RTQ7jrIqDuub95USs38l$|?HlLgwRF7Du=fRin}0-TTfkSy(If
zLwq50G8c_De!mF~jZ)VT5Mx!H$GdIooOjPo;>-jQvF4TYO4^xBE^S}jbUA)q#no>$
zY%#VuAIfC4a>&9pTKS>NWy80r($e)4c5Ab!o$}Ic<b0{o4<DG;Jv2;2m2-T==|fU|
zL2tgYyBIN~`yX|_!T15%^5K2NH983^xeYe@2N$NTgN?4PIJWL9y&<1**Q<j2_q5uX
zGv{T4H^y<r4{W$;{g!EMI7mxxHCsM(F_iO!r+%Vh!^y(brLg;(;r;az!ePpJ=t0c#
z{G4^q>CVb>?l9pkYOW7`LJU}Wz|sK;3H^IIR@d7}OI7%B`Kjm<h}k$X4m6?|8<WM-
z6j9gXh^Qv9ahHJDx3azQV2+a~{lgEP9F8tiHC|1pEa8O~4xYEVbF(RhZ>@Y)UTc+=
zm8`>d$DnqyFm`vU&EkMSJ5CCV_X}U?Cot5sZ%gZx!4E%5>Nl`3y1X$u=j(Itn>@=3
mQ&py3=+xL)Egwctts@eZ^<z|DW<LZW;OSaa(!n&24EqPq1?PnT

literal 0
HcmV?d00001

diff --git a/WGS-processing/cwl/helper/bwa-gatk-report-wf.cwl b/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl
similarity index 100%
rename from WGS-processing/cwl/helper/bwa-gatk-report-wf.cwl
rename to WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl
diff --git a/WGS-processing/cwl/helper/getgvcfs.cwl b/WGS-processing/cwl/helper/not-in-use/getgvcfs.cwl
similarity index 100%
rename from WGS-processing/cwl/helper/getgvcfs.cwl
rename to WGS-processing/cwl/helper/not-in-use/getgvcfs.cwl
diff --git a/WGS-processing/cwl/wgs-processing-wf.cwl b/WGS-processing/cwl/wgs-processing-wf.cwl
index 0a4b9d2..f7d2d00 100644
--- a/WGS-processing/cwl/wgs-processing-wf.cwl
+++ b/WGS-processing/cwl/wgs-processing-wf.cwl
@@ -3,10 +3,10 @@ class: Workflow
 
 requirements:
   - class: SubworkflowFeatureRequirement
+  - class: ScatterFeatureRequirement
 
 inputs:
-  fastq1: File
-  fastq2: File
+  fastqdir: Directory 
   reference:
     type: File
     secondaryFiles:
@@ -29,65 +29,33 @@ inputs:
   tailhtml: File
 
 outputs:
-  qc-html:
-    type: File[]
-    outputSource: fastqc/out-html
-  qc-zip:
-    type: File[]
-    outputSource: fastqc/out-zip 
   gvcf:
-    type: File
-    outputSource: haplotypecaller/gatheredgvcf
+    type: File[]
+    outputSource: bwamem-gatk-report/gvcf
   report:
-    type: File  
-    outputSource: generate-report/report
+    type: File[]  
+    outputSource: bwamem-gatk-report/report
+
 steps:
-  fastqc:
-    run: fastqc.cwl
-    in:
-      fastq1: fastq1
-      fastq2: fastq2
-    out: [out-html, out-zip]
-  bwamem-samtools-view:
-    run: bwamem-samtools-view.cwl
-    in:
-      fastq1: fastq1
-      fastq2: fastq2
-      reference: reference
-      sample: sample
-    out: [bam]
-  samtools-sort:
-    run: samtools-sort.cwl 
-    in:
-      bam: bwamem-samtools-view/bam
-      sample: sample
-    out: [sortedbam]
-  mark-duplicates:
-    run: mark-duplicates.cwl
+  getfastq:
+    run: ./helper/getfastq.cwl
     in:
-      bam: samtools-sort/sortedbam
-    out: [dupbam,dupmetrics]
-  samtools-index:
-    run: samtools-index.cwl
-    in:
-      bam: mark-duplicates/dupbam
-    out: [indexedbam]
-  haplotypecaller:
-    run: scatter-gatk-wf-with-interval.cwl 
+      fastqdir: fastqdir
+    out: [fastq1, fastq2]
+
+  bwamem-gatk-report:
+    run: ./helper/bwamem-gatk-report-wf.cwl
+    scatter: [fastq1, fastq2]
+    scatterMethod: dotproduct
     in:
+      fastq1: getfastq/fastq1
+      fastq2: getfastq/fastq2
       reference: reference
-      bam: samtools-index/indexedbam
       sample: sample
+      knownsites: knownsites
       scattercount: scattercount
-      knownsites1: knownsites
-    out: [gatheredgvcf]
-  generate-report:
-    run: report-wf.cwl
-    in:
-      gvcf: haplotypecaller/gatheredgvcf
-      samplename: sample
       clinvarvcf: clinvarvcf
       reportfunc: reportfunc
       headhtml: headhtml
       tailhtml: tailhtml
-    out: [report]  
+    out: [qc-html,qc-zip,gvcf,report]
-- 
2.30.2