From: Sarah Wait Zaranek <swz@curii.com>
Date: Wed, 24 Jun 2020 23:36:02 +0000 (+0000)
Subject: Adding more labels/metadata
X-Git-Url: https://git.arvados.org/arvados-tutorial.git/commitdiff_plain/b50ffeeeae865d0780ca03c5a3f5065eeeb758e4

Adding more labels/metadata
Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek <swz@curii.com>
no issue #
---

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 0000000..8279731
Binary files /dev/null and b/WGS-processing/cwl/helper/.samtools-sort.cwl.swp differ
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