Upeate structure, dockerize kallisto_index

.gitattributes

@@ -5,4 +5,4 @@
 /assays/Talinum_RNASeq_minimal/dataset/DB_103_CAMMD_AGTCAA_L001_R1_001.fastq.gz filter=lfs diff=lfs merge=lfs -text
 /assays/Talinum_RNASeq_minimal/dataset/DB_161_reC3MD_GTCCGC_L001_R1_001.fastq.gz filter=lfs diff=lfs merge=lfs -text
 /assays/Talinum_RNASeq_minimal/dataset/DB_165_re-C3MD_GTGAAA_L002_R1_001.fastq.gz filter=lfs diff=lfs merge=lfs -text
-/runs/kallisto_index/kallisto_index filter=lfs diff=lfs merge=lfs -text
+/runs/kallisto_index/out/kallisto_index filter=lfs diff=lfs merge=lfs -text

runs/kallisto_index/README.md

@@ -8,14 +8,6 @@ cd /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq/runs/kallisto_index
 ## 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" kallisto_index.yml
-
 ### run with cwltool
-cwltool ../../workflows/kallisto_index.cwl kallisto_index.yml
-
+cwltool --enable-dev run.cwl kallisto_index.yml
 ```

runs/kallisto_index/kallisto_index.yml

 cores: 1
-sh_script:
+in_genome_ref:
   class: File
-  path: ../../workflows/kallisto_index.sh
-in_genome_ref: studies/TalinumGenomeDraft/resources/Talinum.gm.CDS.nt.fa
-out_folder: runs/kallisto_index
-arc_root: /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq
+  path: ../../studies/TalinumGenomeDraft/resources/Talinum.gm.CDS.nt.fa
+out_folder: ./out

runs/kallisto_index/run.cwl

+#!/usr/bin/env cwl-runner
+cwlVersion: v1.2.0-dev1
+class: Workflow
+inputs:
+  in_genome_ref:
+    type: File
+  out_folder:
+    type: string
+outputs:
+  out_dir:
+    type:
+      type: array
+      items: Directory
+    outputSource: kallisto_index/outdir
+steps:
+  kallisto_index:
+    run: ../../workflows/kallisto_index/workflow.cwl
+    in:
+      in_genome_ref: in_genome_ref
+      out_folder: out_folder
+    out: [outdir]
\ No newline at end of file

workflows/kallisto_index.sh → workflows/kallisto_index/kallisto_index.sh

@@ -16,9 +16,8 @@
 #### Read arguments from CLI
 ################################################
 
-arc_root=$1
-in_genome_ref=$2
-out_folder=$3
+in_genome_ref=$1
+out_folder=$2
 
 ################################################
 #### Print version and citation to test kallisto 
@@ -31,11 +30,10 @@ kallisto cite
 #### If it does not exist, create out dir
 ################################################
 
-mkdir -p "$arc_root/$out_folder/"
-
+mkdir -p $out_folder
 
 ################################################
 #### Build kallisto index
 ################################################
 
-kallisto index -i "$arc_root/$out_folder/kallisto_index" "$arc_root/$in_genome_ref"
+kallisto index -i $out_folder/kallisto_index $in_genome_ref

workflows/kallisto_index.cwl → workflows/kallisto_index/workflow.cwl

 #!/usr/bin/env cwl-runner
 
-cwlVersion: v1.2
+cwlVersion: v1.2.0-dev1
 class: CommandLineTool
-
+hints:
+  DockerRequirement:
+    dockerPull: zlskidmore/kallisto:0.48.0
+requirements:
+  - class: InitialWorkDirRequirement
+    listing:
+      - class: File
+        location: kallisto_index.sh
+arguments:
+  - position: 0
+    valueFrom: ./kallisto_index.sh
 inputs:
-- id: sh_script
+- id: in_genome_ref
   type: File
-  inputBinding:
-    position: 0
-- id: arc_root
-  type: string
   inputBinding:
     position: 1
-- id: in_genome_ref
-  type: string
-  inputBinding:
-    position: 2
 - id: out_folder
   type: string
   inputBinding:
-    position: 3
+    position: 2
 
 outputs:
 - id: outdir