Add subsampling module

Dockerfile

@@ -18,7 +18,9 @@ RUN apt-get update && \
     zlib1g-dev \
     libgomp1 \
     procps \
-    libx11-6
+    libx11-6 \
+    bc
+
 RUN apt-get clean && rm -rf /var/lib/apt/lists/*
 
 # Install mutserve (not as conda package available)

cloudgene.yaml

@@ -105,6 +105,12 @@ workflow:
         on: On
         off: Off
 
+    - id: subsampling
+      description: Subsample to specified coverage (Deactived with value 0)
+      type: number
+      visible: true 
+      value: 0  
+
     - id: myseparator0
       type: separator
 

environment.yml

@@ -5,6 +5,7 @@ channels:
   - conda-forge
   - bioconda
 dependencies:
+  - conda-forge::ncurses
   - unzip=6.0
   - openjdk=17
   - r-base=4.3.2

modules/local/subsampling.nf

@@ -0,0 +1,28 @@
+process SUBSAMPLING {
+
+    input:
+    path bam_file
+    val coverage
+
+    output:
+    path "${bam_file}", includeInputs: true, emit: subsampled_bam_ch
+
+    script:
+    def avail_mem = 1024
+    if (task.memory) {
+        avail_mem = (task.memory.mega*0.8).intValue()
+    }    
+
+    """
+    samtools coverage ${bam_file} > samtools_coverage_${bam_file.baseName}.txt
+    csvtk grep -t -f3 -p 16569 -C '\$' samtools_coverage_${bam_file.baseName}.txt
+    mean_cov=\$(csvtk grep -t -f3 -p 16569 -C '\$' samtools_coverage_${bam_file.baseName}.txt | csvtk cut -t -f 7)
+    mean_cov_int=\$(printf "%.0f" "\$mean_cov")
+    fraction=\$(echo "scale=4; 1+(${coverage} / \${mean_cov})" | bc)
+    if [ \${mean_cov_int} -gt ${coverage} ]
+    then
+        samtools view -s \$fraction -b -o ${bam_file.baseName}.subsampled.bam ${bam_file}
+        mv ${bam_file.baseName}.subsampled.bam ${bam_file}
+    fi 
+    """
+}

nextflow.config

@@ -23,6 +23,7 @@ params {
     alignQ                 = 30
     coverage_estimation    = "on"
     max_samples            = 0
+    subsampling            = 0
 
 
     service = [

workflows/mtdna_server_2.nf

@@ -28,6 +28,7 @@ include { ANNOTATE } from '../modules/local/annotate'
 include { HAPLOGROUPS_CONTAMINATION } from '../modules/local/haplogroups_contamination'
 include { COVERAGE_ESTIMATION } from '../modules/local/coverage_estimation'
 include { REPORT } from '../modules/local/report'
+include { SUBSAMPLING } from '../modules/local/subsampling'
 include { SAMPLE_REPORT } from '../modules/local/sample_report'
 
 
@@ -83,6 +84,14 @@ workflow MTDNA_SERVER_2 {
 
     validated_files = INPUT_VALIDATION.out.validated_files.flatten()
 
+    if(params.subsampling != 0) {
+        SUBSAMPLING (
+            validated_files,
+            params.subsampling
+        )
+        validated_files = SUBSAMPLING.out.subsampled_bam_ch
+    }
+
     if (params.mode == 'mutserve') {
 
         MUTSERVE(