From b8ecb14eae00f705b278338e92cb251110a21ace Mon Sep 17 00:00:00 2001
From: Sarah Wait Zaranek <swz@curii.com>
Date: Fri, 26 Jun 2020 01:52:47 +0000
Subject: [PATCH] Adding labels and other metadata
 Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek <swz@curii.com> no issue #

---
 .../cwl/helper/.gatk-splitintervals.cwl.swp   | Bin 0 -> 12288 bytes
 .../gatk-baserecalibrator-with-interval.cwl   |  34 ++++++++++++++----
 .../cwl/helper/gatk-wf-with-interval.cwl      |  13 ++++++-
 3 files changed, 40 insertions(+), 7 deletions(-)
 create mode 100644 WGS-processing/cwl/helper/.gatk-splitintervals.cwl.swp

diff --git a/WGS-processing/cwl/helper/.gatk-splitintervals.cwl.swp b/WGS-processing/cwl/helper/.gatk-splitintervals.cwl.swp
new file mode 100644
index 0000000000000000000000000000000000000000..c7eb3dc46593f011977983b53f26281d0aacd266
GIT binary patch
literal 12288
zcmeI2KWrRD6voHU5Fmg;goK1(a*4!0?e3jpVrMJPc4AJ6ZFF~b5EeP+cIWN}@9iuz
zbG|rD%0NW{G*l_55C{pnNFY&Bfdmy21)_rn5E23^_;z+LXFEhGs6aE)&+FOuX7<fD
zZ^nx9y4B6bDxLS|2-*XLY#sUS;+<bVeC55PGek4R2KoEjv_$j=CKYMli}_B?*8`p;
zz8;;PO?;+ySS&TmgddS0W`=n>j)Sb91*OR|BO{eyNpNAM<z-4nTx-k~MB^mDOc_@}
zpP6k>XNfQ(H5^kW+K2fzVL7u?z$vg#fy3nFY^|EVI$k+OA3U<L&+0l(0jGddz$xGq
za0)mDoB~b(r@;SP0h7&;D_G(k<ucEe&zS?yvg;mB0jGddz$xGqa0)mDoB~b(r+`zy
zDc}@v3j7BZU@;-9cM<aTAq0>A|DXK(|L6M&`38IpJ_7H8H$VisU<=g2qu>zu^*%zr
z2Va5f;C*lnTm*ISELZ?fgGa#chY9%>d=9RI55P6>Cb$YNgKh9KSO*K>aqtlM@m@lH
z0N;Tx!6)Dya20HU7r`Q^gHzxbI0|OKFZU4g1^5uW4hEnK9s>mYbvGfOf!DxQa2a&L
zBB+2rq0=AWC-5oAb$biA<`i%WI0c*nP64NYQ{Z1IKs`DzCJTdr$80E5BNN%*^`+_u
zXO`=$otj@;@FfZ$OGgb)9S!)9S)U!Qw2*0#FvGQ(SXA62Ckihe5<*cxX{1a-S}DI3
z$sxWqPV5||^>~F)6lzYL%BszYPj1g+3~w1Qp}H_02$Gyg*^ujuq23eW=r9oxPc@%9
zkH=%bpN{OON}~=~KTEume%0rxKQO}t10xypCeNf6C?wsTj_(UI7<GLtYf@`EdETfw
zg$!t@Cmr##N@5k2Nr7okBwS;0P&iY(CoaL}t>-WH`!Ckd*VmRBYb%`#jrMcv8|_YG
zt$n7sSwG)tZ7eP~HXE(R`r5&XJU!~hVn^iL<P8!3O%tDZLZsQqXkSOnpyoA_qtrZk
zARezYde(aDIj{h(cpCA86Eus!T6t-0V^UgcKT`2t7+#uNDeM)`dxh;VPi6)ILl&m`
zqajaCzH?os`4UoxrEy4YsjmG|NK0uyk==v322c4DojZxprifgQ3|71-(ir8xA)=r(
zVO@?4Z50~O$skSb_E;CQn;~b;V!O6p!i$NEbqGl@r_J~2bjw@|#batKifK<OWB@9V
zQ$r(`B>4EAiM3J1i@p00QtL5^jt#Sf(>0{?KMu6!a!8&}7yVd7#=7QRzsJP1quKO#
zoQ-LxJKi&drRmhdhF#myJd$b5)NU>YmQ2yg5Sx<(pnaaoGEOXf$V^zA-=fp0ot`~S
zkJGdDR=aK|n97UpjXVlTk)8QHQnvEa3F#KC%=tBczC!c;D4NcAglhn%{I1Dw0udv9
zL{H$VvIkr#5eLm?V>xUsH9OU6ZDIb@bcU6B`&_XgXtK@v#qwIRXU0t7@V`33hUN)b
zI7L!%y(-d>&e{+Z8`|D8)oK+lnq13~if$ga&3V>|7;e&#)~d5}c5snG%R;qnrdglj
zT_+X_F8Te~#OV}6oP!hz!;pl~e^5N^B(aRPxq5z-Ae*~NGNiaR0;TWV9+uE5=I6mJ
zThDz<Qq%S2nOyGalH$1EX_=Qd62CkeQNlD7FAwH<k@7Y!dxC3yv-s-}Qk3e~$ln;f
BYNr4I

literal 0
HcmV?d00001

diff --git a/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl b/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
index dec3244..71cb27f 100644
--- a/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
+++ b/WGS-processing/cwl/helper/gatk-baserecalibrator-with-interval.cwl
@@ -1,10 +1,6 @@
 cwlVersion: v1.1
 class: CommandLineTool
-label: Generating recalibration table for BQSR 
-
-$namespaces:
-  arv: "http://arvados.org/cwl#"
-  cwltool: "http://commonwl.org/cwltool#"
+label: Generate recalibration table for BQSR 
 
 requirements:
   DockerRequirement:
@@ -20,14 +16,23 @@ hints:
   ResourceRequirement:
     ramMin: 5000
     coresMin: 2
+  SoftwareRequirement:
+    packages:
+      GATK:
+        specs: [ "https://identifiers.org/rrid/RRID:SCR_001876" ]
+        version: [ "4.1.7" ]
 
 inputs:
   bam:
     type: File
+    format: edam:format_2572 # BAM
+    label: Indexed sorted BAM with labeled duplicates
     secondaryFiles:
      - .bai
   reference:
     type: File
+    format: edam:format_1929 # FASTA
+    label: Reference genome
     secondaryFiles:
       - .amb
       - .ann
@@ -36,17 +41,23 @@ inputs:
       - .sa
       - .fai
       - ^.dict
-  sample: string
+  sample:
+    type: string
+    label: Sample Name
   knownsites1: 
     type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
     secondaryFiles:
       - .tbi
   intervallist:
     type: File
+    label: Scatter intervals file
 
 outputs:
   recaltable:
     type: File
+    label: Recalibration table
     outputBinding:
       glob: "*.table"
 
@@ -66,3 +77,14 @@ arguments:
     valueFrom: $(inputs.intervallist)
   - prefix: "-O"
     valueFrom: $(inputs.sample)_recal_data.table
+
+s:codeRepository: https://github.com/arvados/arvados-tutorial
+s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
+
+$namespaces:
+ s: https://schema.org/
+ edam: http://edamontology.org/
+
+$schemas:
+ - https://schema.org/version/latest/schema.rdf
+ - http://edamontology.org/EDAM_1.18.owl
diff --git a/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl b/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
index e1e54ee..ef90209 100644
--- a/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
+++ b/WGS-processing/cwl/helper/gatk-wf-with-interval.cwl
@@ -8,10 +8,14 @@ requirements:
 inputs:
   bam:
     type: File
+    format: edam:format_2572 # BAM
+    label: Indexed sorted BAM with labeled duplicates
     secondaryFiles:
       - .bai
   reference:
     type: File
+    format: edam:format_1929 # FASTA
+    label: Reference genome
     secondaryFiles:
       - .amb
       - .ann
@@ -20,17 +24,24 @@ inputs:
       - .sa
       - .fai
       - ^.dict
-  sample: string
+  sample: 
+    type: string
+    label: Sample Name
   knownsites1:
     type: File
+    format: edam:format_3016 # VCF
+    label: VCF of known polymorphic sites for BQSR
     secondaryFiles:
       - .tbi
   intervallist:
     type: File
+    label: Scatter intervals file 
 
 outputs:
   gvcf:
     type: File
+    format: edam:format_3016 # GVCF
+    label: single interval filtered GVCF
     outputSource: selectvariants/filteredgvcf
 
 steps:
-- 
2.30.2