Dockerize merge_isa_metadata

runs/merged_isa_metadata/README.md

@@ -9,15 +9,6 @@ cd /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq/runs/merged_isa_meta
 ## Let it flow
 
 ```bash
-
-### store arc root (two levels up from here) as variable
-arc_root=$(echo ${PWD%/*/*})
-
-### replace arc root line in yml (specific to the machine from where this is run)
-### not sure, if this works on linux... 
-sed -i '' "s|^arc_root:.*|arc_root: $arc_root|g" merge_isa_metadata.yml
-
 ### run with cwltool
-cwltool ../../workflows/merge_isa_metadata.cwl merge_isa_metadata.yml
-
+cwltool --enable-dev run.cwl merge_isa_metadata.yml
 ```

runs/merged_isa_metadata/merge_isa_metadata.yml

 cores: 1
-r_script:
+in_isa_study: 
   class: File
-  path: ../../workflows/merge_isa_metadata.R
-in_isa_study: studies/TalinumFacultativeCAM/isa.study.xlsx:plant_growth
-in_isa_assay: assays/Talinum_RNASeq_minimal/isa.assay.xlsx:2EXT01_RNA:3ASY01_RNASeq
-out_folder: runs/merged_isa_metadata
-arc_root: /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq
+  path: ../../studies/TalinumFacultativeCAM/isa.study.xlsx
+in_isa_assay: 
+  class: File
+  path: ../../assays/Talinum_RNASeq_minimal/isa.assay.xlsx
+out_folder: out
\ No newline at end of file

runs/merged_isa_metadata/run.cwl

+#!/usr/bin/env cwl-runner
+cwlVersion: v1.2.0-dev1
+class: Workflow
+inputs:
+  in_isa_study:
+    type: File
+  in_isa_assay:
+    type: File
+  out_folder:
+    type: string
+outputs:
+  out_dir:
+    type:
+      type: array
+      items: Directory
+    outputSource: merge_isa_metadata/outdir
+steps:
+  merge_isa_metadata:
+    run: ../../workflows/merge_isa_metadata/workflow.cwl
+    in:
+      in_isa_study: in_isa_study
+      in_isa_assay: in_isa_assay
+      out_folder: out_folder
+    out: [outdir]

workflows/merge_isa_metadata/merge_isa_metadata.R

@@ -21,17 +21,18 @@ library(openxlsx)
 
 args <- commandArgs(trailingOnly = T)
 
-arc_root <- args[1]
-in_isa_study <- args[2]
-in_isa_assay <- args[3]
-out_folder <- args[4]
+in_isa_study <- paste(args[1], ":plant_growth", sep="")
+print(in_isa_study)
+in_isa_assay <- paste(args[2], ":2EXT01_RNA:3ASY01_RNASeq", sep="")
+print(in_isa_assay)
+out_folder <- args[3]
 
 ################################################
 #### Read metadata from isa excel workbooks
 ################################################
 
-isa_study <- unlist(strsplit(paste(arc_root, in_isa_study, sep = "/"), split = ":"))
-isa_assay <- unlist(strsplit(paste(arc_root, in_isa_assay, sep = "/"), split = ":"))
+isa_study <- unlist(strsplit(in_isa_study, split = ":"))
+isa_assay <- unlist(strsplit(in_isa_assay, split = ":"))
 
 isa_sheets <- list()
 for (i in 2:length(isa_study))
@@ -76,12 +77,10 @@ isa_merged <- isa_merged[, !apply(isa_merged, 2, function(x) {
 #### If it does not exist, create out dir
 ################################################
 
-
-dir.create(paste(arc_root, out_folder, sep = "/"), recursive = T, showWarnings = F)
-
+dir.create(out_folder, recursive = T, showWarnings = F)
 
 ################################################
 #### Write table to file
 ################################################
 
-write.table(isa_merged, file = paste(arc_root, out_folder, "merged_isa.tsv", sep = "/"), sep = "\t")
+write.table(isa_merged, file = paste(out_folder, "merged_isa.tsv", sep = "/"), sep = "\t")

workflows/merge_isa_metadata/workflow.cwl

 #!/usr/bin/env cwl-runner
 
-cwlVersion: v1.2
+cwlVersion: v1.2.0-dev1
 class: CommandLineTool
-
+hints:
+  DockerRequirement:
+    dockerPull: zimmera95/rnaseq:latest
+requirements:
+  - class: InitialWorkDirRequirement
+    listing:
+      - class: File
+        location: merge_isa_metadata.R
+arguments:
+  - position: 0
+    valueFrom: merge_isa_metadata.R
 inputs:
-- id: r_script
+- id: in_isa_study
   type: File
-  inputBinding:
-    position: 0
-- id: arc_root
-  type: string
   inputBinding:
     position: 1
-- id: in_isa_study
-  type: string
-  inputBinding:
-    position: 2
 - id: in_isa_assay
-  type: string
+  type: File
   inputBinding:
-    position: 3
+    position: 2
 - id: out_folder
   type: string
   inputBinding:
-    position: 4
+    position: 3
 
 outputs:
 - id: outdir