Merge pull request #3 from genomic-medicine-sweden/chewbbaca-batch-input

Chewbbaca batch input and file lock removal
genomic-medicine-sweden · Feb 9, 2023 · 823ea22 · 823ea22
2 parents 16f144f + 3d7a862
commit 823ea22
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 23 deletions.
diff --git a/modules/bwa/main.nf b/modules/bwa/main.nf
@@ -23,7 +23,7 @@ process bwa_index {
 
   script:
     """
-    bwa index ${reference}
+    bwa index ${reference} ${reference.baseName}/${reference}
     """
 }
 

diff --git a/modules/chewbbaca/main.nf b/modules/chewbbaca/main.nf
@@ -13,60 +13,76 @@ params = initParams(params)
 
 process chewbbaca_allelecall {
   label "process_medium"
-  tag "${sampleName}"
+  tag "${workflow.runName}"
   publishDir "${params.publishDir}", 
     mode: params.publishDirMode, 
     overwrite: params.publishDirOverwrite
 
   input:
-    tuple val(sampleName), path(input)
+    val sampleName
+    path batchInput
     path schemaDir
     path trainingFile
 
   output:
-    tuple val(sampleName), path('output_dir/*/results_alleles.tsv')
+    val sampleName, emit: sampleName
+    path 'output_dir/results_alleles.tsv', emit: calls
     //tuple val(sampleName), path("${missingLoci}"), emit: missing
 
   script:
-    output = "${sampleName}.vcf"
     missingLoci = "chewbbaca.missingloci"
     trainingFile = trainingFile ? "--ptf ${trainingFile}" : "" 
-    flockfile = file(params.localTempDir + '/chewbbaca.lock') 
-    flocking = flockfile.exists() ? "flock -e $flockfile \\" : ""
 
     """
-    echo ${input} > batch_input.list
-    ${flocking}
-      chewBBACA.py AlleleCall \\
-      -i batch_input.list \\
-      ${params.args.join(' ')} \\
-      --cpu ${task.cpus} \\
-      --output-directory output_dir \\
-      ${trainingFile} \\
-      --schema-directory ${schemaDir}
+    chewie AlleleCall \\
+    -i ${batchInput} \\
+    ${params.args.join(' ')} \\
+    --cpu ${task.cpus} \\
+    --output-directory output_dir \\
+    ${trainingFile} \\
+    --schema-directory ${schemaDir}
     #bash parse_missing_loci.sh batch_input.list 'output_dir/*/results_alleles.tsv' ${missingLoci}
     """
 }
 
+process chewbbaca_create_batch_list {
+  label "process_low"
+  publishDir params.publishDir, 
+    mode: params.publishDirMode, 
+    overwrite: params.publishDirOverwrite
+
+  input:
+    path maskedAssembly
+
+  output:
+    path "batch_input.list"
+
+  script:
+    output = "batch_input.list"
+    """
+    realpath $maskedAssembly > $output
+    """
+}
+
 process chewbbaca_split_results {
   label "process_low"
-  tag "${assembly.simpleName}"
+  tag "${sampleName}"
   publishDir "${params.publishDir}", 
     mode: params.publishDirMode, 
     overwrite: params.publishDirOverwrite
 
   input:
+    each sampleName
     path input
 
   output:
-    path("${output}")
+    tuple val(sampleName), path("${output}")
 
   script:
-    id = "${input.simpleName}"
-    output = "${id}.chewbbaca"
+    output = "${sampleName}.chewbbaca"
     """
     head -1 ${input} > ${output}
-    grep ${id} ${input} >> ${output}
+    grep ${sampleName} ${input} >> ${output}
     """
 }
 

diff --git a/modules/cmd/main.nf b/modules/cmd/main.nf
@@ -156,8 +156,6 @@ process save_analysis_metadata {
     mode: params.publishDirMode, 
     overwrite: params.publishDirOverwrite
 
-  input:
-
   output:
     path(output)