Updates, CI, config, main, check_design

.gitlab-ci.yml

 before_script:
   - module load python/3.6.1-2-anaconda
-  - module load nextflow/0.27.6
-  - ln -s /project/shared/bicf_workflow_ref/workflow_testdata/cellranger_mkfastq/*tar.gz test_data/
+  - module load nextflow/0.31.1_Ignite
+  - mkdir test_data/simple
+  - ln -s /project/shared/bicf_workflow_ref/workflow_testdata/cellranger/cellranger_mkfastq/simple/* test_data/simple/
 
 stages:
-  - integration
+  - astrocyte
+  - simple
+
+astrocyte_check:
+  stage: astrocyte
+  script:
+  - astrocyte_cli check ../cellranger_mkfastq
 
 simple_test:
-  stage: integration
+  stage: simple
   script:
-  - nextflow run workflow/main.nf
+  - nextflow run workflow/main.nf --bcl test_data/simple/*.tar.gz --designFile test_data/simple/cellranger-tiny-bcl-simple-1_2_0.csv

workflow/conf/biohpc.config

@@ -8,6 +8,7 @@ process {
     executor = 'local'
   }
   $untarBCL {
+    module = ['pigz/2.4']
     queue = 'super'
   }
   $mkfastq {

workflow/main.nf

@@ -6,14 +6,16 @@
 // Define Input variables
 params.bcl = "$baseDir/../test_data/*.tar.gz"
 params.designFile = "$baseDir/../test_data/design.csv"
-
+params.outDir = "$baseDir/output"
 
 // Define List of Files
 tarList = Channel.fromPath( params.bcl )
 
-
 // Define regular variables
-
+designLocation = Channel
+  .fromPath(params.designFile)
+  .ifEmpty { exit 1, "design file not found: ${params.designFile}" }
+outDir = params.outDir
 
 process checkDesignFile {
 
@@ -21,11 +23,11 @@ process checkDesignFile {
 
   input:
 
-  params.designFile
+  file designLocation
 
   output:
 
-  file("design.csv") into designPaths
+  file("design.checked.csv") into designPaths
 
   script:
 
@@ -33,7 +35,7 @@ process checkDesignFile {
   hostname
   ulimit -a
   module load python/3.6.1-2-anaconda
-  python3 $baseDir/scripts/check_design.py -d $params.designFile
+  python3 $baseDir/scripts/check_design.py -d $designLocation
   """
 }
 
@@ -65,7 +67,7 @@ process untarBCL {
 process mkfastq {
   tag "${bcl.baseName}"
 
-  publishDir "$outDir/${task.process}/${bcl.baseName}", mode: 'copy'
+  publishDir "$outDir/${task.process}", mode: 'copy'
 
   input:
 

workflow/scripts/check_design.py

@@ -70,7 +70,7 @@ def main():
     # Check design file
     new_design_df = check_design_headers(design_df)
 
-    new_design_df.to_csv('design.csv', header=True, sep=',', index=False)
+    new_design_df.to_csv('design.checked.csv', header=True, sep=',', index=False)
 
 if __name__ == '__main__':
     main()