From b50ffeeeae865d0780ca03c5a3f5065eeeb758e4 Mon Sep 17 00:00:00 2001
From: Sarah Wait Zaranek <swz@curii.com>
Date: Wed, 24 Jun 2020 23:36:02 +0000
Subject: [PATCH 1/1] Adding more labels/metadata
 Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek <swz@curii.com> no issue #

---
 .../cwl/helper/.samtools-sort.cwl.swp         | Bin 0 -> 12288 bytes
 WGS-processing/cwl/helper/mark-duplicates.cwl |  30 ++++++++++++---
 WGS-processing/cwl/helper/samtools-sort.cwl   |  35 ++++++++++++++----
 3 files changed, 53 insertions(+), 12 deletions(-)
 create mode 100644 WGS-processing/cwl/helper/.samtools-sort.cwl.swp

diff --git a/WGS-processing/cwl/helper/.samtools-sort.cwl.swp b/WGS-processing/cwl/helper/.samtools-sort.cwl.swp
new file mode 100644
index 0000000000000000000000000000000000000000..82797313cf62ffcb01fb2983f75a26dc2d75c951
GIT binary patch
literal 12288
zcmeI2&x;&I6vt}?BciKfPX0Jlb{AsQ?w*-!Owu5_KQt`5BC~5eNHWxP*YuRz-PNh8
z-eHlDyMiF`An3)L;w9o$NKXC(dJwz_dQvcg7jyGl)jd1A3lcmUs^CLW^<GuI`qt}#
zW#6!KabuI74VMYpV}$Jd{MY3hH=ek*`r#bWOtG<@ze~$SHa4XwvmoWW-B6Eto`<?V
z-^)X$c3CPlt3;TPC}oBPI!&W0tD@EvnA%7sSRTE7VJoPVOt{wAD@rDLgq<p`BFzdT
zWv&A)l?h?EOF-W274QliR-iVc;M_bp-RpL&VR7LEJ^uJxhkg9WE8rFI3U~#)0$u^H
zfLFjP;1#&16)@EtxrQ6w*WU7_wx2uFw?p4~1-t@Y0k42pz$@St@CtYZyaHYUuYgy;
zEAT&5fTe`|@gO0`9zyW=|Nr#w|J#oe@*}tfZi27DXF!2%@FI8)JO!Qv$HARP2>Auv
z0$+kp!3SUi1mJnF1fB--;J0}~z60NaufWIPI(Q#k1toY5EQ1r^0dNfb`7j~3!LQ&a
z@B_FBz5t(tPrygu3aCI2+`;|+2ET*v!8hPKcozsT01>b{+z)*73U~#)0$u^HfLFjP
z;1#%+6`%nf8&kzm#8Xzt(#Tw9d!bZW^ve3m=AavP&xI0or>&!gr%uMaV9sZkD=lOh
z<;-wxrWTcs$Wr5_V?rqER&5n)Oc&a-5?SD^N$&P&t&0nUqB3^o)RxIkeXQNa82@X)
zgzCn4BuKJUvV!Z1p}-d5=r|V%FEyVzPbQNvD{FUDYt#|TsysLyc0yi;V^ic<n8=j(
zc_p<#<?h~WeJ0Gf9)`Hqw77TL?-%((21vN{j3}$x=n#dyRnAX^wyu|_;gA*19!S@n
z>|Hq7jGO8GW$+6tX2z%5-Beu~QE;5BxNatcmz~XSOqj~*f|tf#d&o3jgYK+MV>+!X
zwo%!Z<#Gt+pPhvRwb5j45bmldQxqo0ZYpO(j;Gvmfz#EMP3opcQWea^%@Z|)LGR4j
z9$mCP&c*Ci(EL@Aa}1l4aG#hohqz3kluq||0?{|{at9V`qY#K)?#zcGT0u2->26C!
z=>luQaLfN|cs4Lfn#Zy!+_pB~WBLwVz^Bn!pqKV-y5b2E5Eh%|HU~ut?ZrsoBXOyv
zlt`m~e`7t~TI&xw9aN;AneDLE)_OOWLbmR(%qA>_)*~}v3aNLc7K+<DHcpLl>q2ci
zX?HlU2F=_c70N~C<`RZaHm2u$-52a@+0U%sWX_r_p(ztCN5)O(*13$@6lF}8-B_{0
zjd~q4%=EdIwMvf0v}H@++cY~EE`9H++a{IC4p(p1IZ`O8l@QTzg39=5nR_J?rc!vW
yJHLZF5H-8K&6R1MAP<oe&3oI_N^++03~dari;{2Sw};?Y=3?_>6;qVzcga5`4)iqu

literal 0
HcmV?d00001

diff --git a/WGS-processing/cwl/helper/mark-duplicates.cwl b/WGS-processing/cwl/helper/mark-duplicates.cwl
index 3d7cba6..6041221 100644
--- a/WGS-processing/cwl/helper/mark-duplicates.cwl
+++ b/WGS-processing/cwl/helper/mark-duplicates.cwl
@@ -1,9 +1,6 @@
 cwlVersion: v1.1
 class: CommandLineTool
-label: Marking Duplicates using Picard 
-$namespaces:
-  arv: "http://arvados.org/cwl#"
-  cwltool: "http://commonwl.org/cwltool#"
+label: Marking Duplicates using GATK (Picard) 
 
 requirements:
   DockerRequirement:
@@ -15,18 +12,28 @@ hints:
     coresMin: 4    
   arv:RuntimeConstraints:
     outputDirType: keep_output_dir
+  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: Sorted BAM 
 
 outputs:
   dupbam:
     type: File
+    format: edam:format_2572 # BAM
+    label: Sorted BAM with labeled duplicates
     outputBinding:
       glob: "*.bam"
   dupmetrics:
     type: File
+    label: Duplication metrics file
     outputBinding:
       glob: "*.txt"
 
@@ -41,4 +48,17 @@ arguments:
   - prefix: "-O"
     valueFrom: marked_dups$(inputs.bam.basename)
   - prefix: "-M"
-    valueFrom: "metrics.txt" 
+    valueFrom: "metrics.txt"
+
+s:codeRepository: https://github.com/arvados/arvados-tutorial
+s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
+
+$namespaces:
+ s: https://schema.org/
+ edam: http://edamontology.org/
+ arv: "http://arvados.org/cwl#"
+ cwltool: "http://commonwl.org/cwltool#"
+
+$schemas:
+ - https://schema.org/version/latest/schema.rdf
+ - http://edamontology.org/EDAM_1.18.owl 
diff --git a/WGS-processing/cwl/helper/samtools-sort.cwl b/WGS-processing/cwl/helper/samtools-sort.cwl
index 02072cd..88950e3 100644
--- a/WGS-processing/cwl/helper/samtools-sort.cwl
+++ b/WGS-processing/cwl/helper/samtools-sort.cwl
@@ -1,10 +1,6 @@
 cwlVersion: v1.1
 class: CommandLineTool
-label: Sorting Bam file
-
-$namespaces:
-  arv: "http://arvados.org/cwl#"
-  cwltool: "http://commonwl.org/cwltool#"
+label: Sorting BAM file
 
 requirements:
   DockerRequirement:
@@ -18,14 +14,26 @@ hints:
   arv:RuntimeConstraints:
     keep_cache: 9216 
     outputDirType: keep_output_dir
+  SoftwareRequirement:
+    packages:
+      Samtools:
+        specs: [ "https://identifiers.org/rrid/RRID:SCR_002105" ]
+        version: [ "1.10" ]
 
 inputs:
-  bam: File
-  sample: string
+  bam: 
+    type: File
+    format: edam:format_2572 # BAM
+    label: Alignments in BAM format
+  sample:
+    type: string
+    label: Sample Name
 
 outputs:
   sortedbam:
     type: File
+    format: edam:format_2572 # BAM
+    label: Sorted BAM 
     outputBinding:
       glob: "*sorted.bam"
 
@@ -40,3 +48,16 @@ arguments:
   - '2G'
   - -o
   - $(inputs.sample).sorted.bam
+
+s:codeRepository: https://github.com/arvados/arvados-tutorial
+s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
+
+$namespaces:
+ s: https://schema.org/
+ edam: http://edamontology.org/
+ arv: "http://arvados.org/cwl#"
+ cwltool: "http://commonwl.org/cwltool#"
+
+$schemas:
+ - https://schema.org/version/latest/schema.rdf
+ - http://edamontology.org/EDAM_1.18.owl
-- 
2.30.2