From cde2abc048201aadf3c9dc9c501182bfed9ac43e Mon Sep 17 00:00:00 2001
From: Adrian Zimmer <z.adrian1995@gmail.com>
Date: Sat, 13 Aug 2022 15:24:05 +0200
Subject: [PATCH] Dockerize Kallisto_sleuth
---
runs/kallisto_quant/README.md | 10 +----
runs/kallisto_quant/kallisto_quant.yml | 2 +-
runs/kallisto_quant/run.cwl | 2 +-
runs/kallisto_sleuth/README.md | 10 +----
runs/kallisto_sleuth/kallisto_sleuth.yml | 8 ++--
runs/kallisto_sleuth/{ => out}/sleuth_dge.csv | 0
runs/kallisto_sleuth/run.cwl | 21 ++++++++++
workflows/kallisto_sleuth/kallisto_sleuth.R | 14 +++----
workflows/kallisto_sleuth/workflow.cwl | 39 ++++++++++---------
9 files changed, 53 insertions(+), 53 deletions(-)
rename runs/kallisto_sleuth/{ => out}/sleuth_dge.csv (100%)
create mode 100644 runs/kallisto_sleuth/run.cwl
diff --git a/runs/kallisto_quant/README.md b/runs/kallisto_quant/README.md
index 7e03991..34ee542 100644
--- a/runs/kallisto_quant/README.md
+++ b/runs/kallisto_quant/README.md
@@ -8,14 +8,6 @@ cd /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq/runs/kallisto_quant
## 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: $arc_root|g" kallisto_quant.yml
-
### run with cwltool
-cwltool ../../workflows/kallisto_quant.cwl kallisto_quant.yml
-
+cwltool --enable-dev run.cwl kallisto_quant.yml
```
diff --git a/runs/kallisto_quant/kallisto_quant.yml b/runs/kallisto_quant/kallisto_quant.yml
index 44f6bfe..e5c7505 100644
--- a/runs/kallisto_quant/kallisto_quant.yml
+++ b/runs/kallisto_quant/kallisto_quant.yml
@@ -9,4 +9,4 @@ in_fastq_dir:
kallisto_bootstrap: 100
kallisto_threads: 4
kallisto_fragmentLength: 200
-kallisto_stdDev: 20
\ No newline at end of file
+kallisto_stdDev: 20
diff --git a/runs/kallisto_quant/run.cwl b/runs/kallisto_quant/run.cwl
index a5e9b88..87cc30f 100644
--- a/runs/kallisto_quant/run.cwl
+++ b/runs/kallisto_quant/run.cwl
@@ -33,4 +33,4 @@ steps:
kallisto_threads: kallisto_threads
kallisto_fragmentLength: kallisto_fragmentLength
kallisto_stdDev: kallisto_stdDev
- out: [outdir]
\ No newline at end of file
+ out: [outdir]
diff --git a/runs/kallisto_sleuth/README.md b/runs/kallisto_sleuth/README.md
index 05639e7..9df2ee4 100644
--- a/runs/kallisto_sleuth/README.md
+++ b/runs/kallisto_sleuth/README.md
@@ -8,14 +8,6 @@ cd /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq/runs/kallisto_sleuth
## Let it flow
```bash
-### store arc root (two levels up from here) as variable
-arc_root=$(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_sleuth.yml
-
### run with cwltool
-cwltool ../../workflows/kallisto_sleuth.cwl kallisto_sleuth.yml
-
+cwltool --enable-dev run.cwl kallisto_sleuth.yml
```
diff --git a/runs/kallisto_sleuth/kallisto_sleuth.yml b/runs/kallisto_sleuth/kallisto_sleuth.yml
index 34364fd..49c1123 100644
--- a/runs/kallisto_sleuth/kallisto_sleuth.yml
+++ b/runs/kallisto_sleuth/kallisto_sleuth.yml
@@ -1,7 +1,5 @@
cores: 1
-r_script:
+in_sleuth:
class: File
- path: ../../workflows/kallisto_sleuth.R
-in_sleuth: runs/kallisto_collect/kallisto_sleuthObject.RData
-out_folder: runs/kallisto_sleuth
-arc_root: /Users/dominikbrilhaus/gitlab_dataplant/samplearc_rnaseq
\ No newline at end of file
+ path: ../kallisto_collect/kallisto_sleuthObject.RData
+out_folder: out
diff --git a/runs/kallisto_sleuth/sleuth_dge.csv b/runs/kallisto_sleuth/out/sleuth_dge.csv
similarity index 100%
rename from runs/kallisto_sleuth/sleuth_dge.csv
rename to runs/kallisto_sleuth/out/sleuth_dge.csv
diff --git a/runs/kallisto_sleuth/run.cwl b/runs/kallisto_sleuth/run.cwl
new file mode 100644
index 0000000..9565052
--- /dev/null
+++ b/runs/kallisto_sleuth/run.cwl
@@ -0,0 +1,21 @@
+#!/usr/bin/env cwl-runner
+cwlVersion: v1.2.0-dev1
+class: Workflow
+inputs:
+ in_sleuth:
+ type: File
+ out_folder:
+ type: string
+outputs:
+ out_dir:
+ type:
+ type: array
+ items: Directory
+ outputSource: kallisto_quant/outdir
+steps:
+ kallisto_quant:
+ run: ../../workflows/kallisto_sleuth/workflow.cwl
+ in:
+ in_sleuth: in_sleuth
+ out_folder: out_folder
+ out: [outdir]
diff --git a/workflows/kallisto_sleuth/kallisto_sleuth.R b/workflows/kallisto_sleuth/kallisto_sleuth.R
index ec3a84a..e4c895e 100644
--- a/workflows/kallisto_sleuth/kallisto_sleuth.R
+++ b/workflows/kallisto_sleuth/kallisto_sleuth.R
@@ -1,5 +1,4 @@
#!/usr/bin/env Rscript
-
################################################
### Diff. gene expression with sleuth ##########
################################################
@@ -15,30 +14,27 @@
################################################
#### Load required library
################################################
-
library(sleuth)
-
################################################
#### Read arguments from CLI
################################################
args <- commandArgs(trailingOnly = T)
-arc_root <- args[1]
-in_sleuth <- args[2]
-out_folder <- args[3]
+in_sleuth <- args[1]
+out_folder <- args[2]
################################################
#### 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)
################################################
#### Load sleuth object
################################################
-load(file = paste(arc_root, in_sleuth, sep = "/"))
+load(file = in_sleuth)
################################################
#### Run sleuth fit
@@ -55,4 +51,4 @@ sleuth_table <- sleuth_results(so, "reduced:full", "lrt", show_all = FALSE)
#### write to file
################################################
-write.csv(sleuth_table, paste(arc_root, out_folder, "sleuth_dge.csv", sep = "/"), row.names = F)
+write.csv(sleuth_table, paste(out_folder, "sleuth_dge.csv", sep = "/"), row.names = F)
diff --git a/workflows/kallisto_sleuth/workflow.cwl b/workflows/kallisto_sleuth/workflow.cwl
index 6f42726..009ea53 100644
--- a/workflows/kallisto_sleuth/workflow.cwl
+++ b/workflows/kallisto_sleuth/workflow.cwl
@@ -1,26 +1,27 @@
#!/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: kallisto_sleuth.R
+arguments:
+ - position: 0
+ valueFrom: kallisto_sleuth.R
inputs:
-- id: r_script
- type: File
- inputBinding:
- position: 0
-- id: arc_root
- type: string
- inputBinding:
- position: 1
-- id: in_sleuth
- type: string
- inputBinding:
- position: 2
-- id: out_folder
- type: string
- inputBinding:
- position: 3
-
+ - id: in_sleuth
+ type: File
+ inputBinding:
+ position: 1
+ - id: out_folder
+ type: string
+ inputBinding:
+ position: 2
outputs:
- id: outdir
type:
--
GitLab