Merge pull request #2 from genepi/features/parallelize-variant-calling

Features/parallelize variant calling

Dockerfile

@@ -38,3 +38,7 @@ RUN jbang export portable -O=PatternSearch.jar PatternSearch.java
 
 COPY ./bin/MutservePerformance.java ./
 RUN jbang export portable -O=MutservePerformance.jar MutservePerformance.java
+
+WORKDIR "/opt"
+ENV GENOMIC_UTILS_VERSION="v0.3.7"
+RUN wget https://github.com/genepi/genomic-utils/releases/download/${GENOMIC_UTILS_VERSION}/genomic-utils.jar

modules/local/call_variants_mutserve.nf

@@ -1,36 +1,36 @@
 process CALL_VARIANTS_MUTSERVE {
 
-  publishDir "${params.outdir}/mutserve", mode: "copy"
+  //publishDir "${params.outdir}/mutserve", mode: "copy"
 
   input:
-  path bamFiles
+  path bam_file
   path ref_fasta
   val(contig)
 
   output:
-  path "${params.project}.txt", emit: variants_ch
-  path "${params.project}_raw.txt", emit: variants_raw_ch
+  path "${bam_file.simpleName}.txt", emit: variants_ch
+  path "${bam_file.simpleName}_raw.txt", emit: variants_raw_ch
 
   script:
   def avail_mem = 1024
   if (!task.memory) {
-      log.info 'Available memory not known - defaulting to 1GB. Specify process memory requirements to change this.'
+      //log.info 'Available memory not known - defaulting to 1GB. Specify process memory requirements to change this.'
   } else {
       avail_mem = (task.memory.mega*0.8).intValue()
   }
 
   """
 	java -Xmx${avail_mem}M -jar /opt/mutserve/mutserve.jar \
       call \
-      --output ${params.project}.vcf \
-      --write-raw --reference ${ref_fasta} \
+      --output ${bam_file.simpleName}.txt \
+      --write-raw \
+      --reference ${ref_fasta} \
       --level ${params.mutserve_detection_limit} \
       --deletions \
       --baseQ ${params.mutserve_baseQ} \
       --mapQ ${params.mutserve_mapQ} \
       --alignQ ${params.mutserve_alignQ} \
       --contig-name ${contig} \
-      --threads ${task.cpus} \
       --no-ansi \$PWD
 	"""
 }

modules/local/detect_type.nf

@@ -1,5 +1,5 @@
 process DETECT_TYPE {
-  publishDir "${params.outdir}/pattern", mode: "copy", pattern: '*pattern.txt'
+  //publishDir "${params.outdir}/pattern", mode: "copy", pattern: '*pattern.txt'
   input:
     path bamFile
     path lpaRegion

modules/local/merge_variants_mutserve.nf

@@ -0,0 +1,27 @@
+process MERGE_VARIANTS_MUTSERVE {
+
+  publishDir "${params.outdir}/variant_calling", mode: "copy"
+
+  input:
+  path variants
+  path variants_raw
+
+  output:
+  path "${params.project}.txt.gz"
+  path "${params.project}_raw.txt.gz"
+  """
+  java -jar /opt/genomic-utils.jar csv-concat \
+      --separator '\t' \
+      --output-sep '\t' \
+      --gz \
+      --output ${params.project}.txt.gz \
+      ${variants}
+
+  java -jar /opt/genomic-utils.jar csv-concat \
+      --separator '\t' \
+      --output-sep '\t' \
+      --gz \
+      --output ${params.project}_raw.txt.gz \
+      ${variants_raw}
+  """
+}

workflows/vntr-calling.nf

@@ -27,6 +27,7 @@ include { EXTRACT_READS         } from '../modules/local/extract_reads'  addPara
 include { BAM_TO_FASTQ          } from '../modules/local/bam_to_fastq'  addParams(outdir: "$outdir")
 include { REALIGN_FASTQ         } from '../modules/local/realign_fastq'  addParams(outdir: "$outdir")
 include { CALL_VARIANTS_MUTSERVE} from '../modules/local/call_variants_mutserve'  addParams(outdir: "$outdir")
+include { MERGE_VARIANTS_MUTSERVE} from '../modules/local/merge_variants_mutserve'  addParams(outdir: "$outdir")
 include { CALCULATE_PERFORMANCE } from '../modules/local/calculate_performance'  addParams(outdir: "$outdir")
 
 
@@ -54,8 +55,9 @@ workflow VNTR_CALLING {
     EXTRACT_READS ( bam_bed_tuple )
     BAM_TO_FASTQ ( EXTRACT_READS.out.extracted_bams_ch )
     REALIGN_FASTQ ( BAM_TO_FASTQ.out.fastq_ch,ref_fasta,BUILD_BWA_INDEX.out.bwa_index_ch )
-    CALL_VARIANTS_MUTSERVE ( REALIGN_FASTQ.out.realigned_ch.collect(),ref_fasta, contig )
-
+    CALL_VARIANTS_MUTSERVE ( REALIGN_FASTQ.out.realigned_ch,ref_fasta, contig )
+    MERGE_VARIANTS_MUTSERVE ( CALL_VARIANTS_MUTSERVE.out.variants_ch.collect(), CALL_VARIANTS_MUTSERVE.out.variants_raw_ch.collect())
+
     if(params.gold != null) {
       gold_standard = file(params.gold, checkIfExists: true)
       CALCULATE_PERFORMANCE ( CALL_VARIANTS_MUTSERVE.out.variants_ch,gold_standard )