Add basic QC calculation

environment.yml

@@ -18,3 +18,5 @@ dependencies:
   - r-rmdformats=1.0.3
   - r-knitr=1.37
   - pandoc=2.14.2
+  - samtools=1.18
+  - csvtk=0.26.0

modules/local/calculate_statistics.nf

@@ -0,0 +1,34 @@
+process CALCULATE_STATISTICS {
+
+    input:
+    path bam_file
+
+    output:
+    path("*summary.txt"), emit: stats_ch
+
+
+    script:
+    def output_name = "${bam_file.baseName}.summary.txt"
+
+    """
+    ## calculate summary statistics
+    samtools coverage ${bam_file} > samtools_coverage_${bam_file.baseName}.txt
+    csvtk grep -t -f3 -p 16569 samtools_coverage_${bam_file.baseName}.txt -T -o mtdna.txt
+
+    contig=\$(csvtk cut -t -f 1 mtdna.txt)
+    numreads=\$(csvtk cut -t -f 4 mtdna.txt)
+    covered_bases=\$(csvtk cut -t -f 5 mtdna.txt)
+    covered_bases_percentage=\$(csvtk cut -t -f 6 mtdna.txt)
+    mean_depth=\$(csvtk cut -t -f 7 mtdna.txt)
+    mean_base_quality=\$(csvtk cut -t -f 8 mtdna.txt)
+    mean_map_quality=\$(csvtk cut -t -f 9 mtdna.txt)
+    echo Sample\tParameter\tValue > $output_name
+    echo ${bam_file.baseName}\tContig\t\${contig} >> $output_name
+    echo ${bam_file.baseName}\tNumberofReads\t\${numreads} >> $output_name
+    echo ${bam_file.baseName}\tCoveredBases\t\${covered_bases} >> $output_name
+    echo ${bam_file.baseName}\tCoveragePercentage\t\${covered_bases_percentage} >> $output_name
+    echo ${bam_file.baseName}\tMeanDepth\t\${mean_depth} >> $output_name
+    echo ${bam_file.baseName}\tMeanBaseQuality\t\${mean_base_quality} >> $output_name
+    echo ${bam_file.baseName}\tMeanMapQuality\t\${mean_map_quality} >> $output_name
+    """
+}

modules/local/summarize_statistics.nf

@@ -0,0 +1,15 @@
+process SUMMARIZE_STATISTICS {
+
+    publishDir "${params.output}/statistics", mode: 'copy'
+
+    input:
+    path statistics
+
+    output:
+    path("sample_statistics.txt"), emit: stats_summarized_ch
+
+
+    """
+    csvtk concat ${statistics} -T -o sample_statistics.txt
+    """
+}

workflows/mitocalling.nf

@@ -1,5 +1,6 @@
 
-
+include { CALCULATE_STATISTICS } from '../modules/local/calculate_statistics'
+include { SUMMARIZE_STATISTICS } from '../modules/local/summarize_statistics'
 include { MUTSERVE } from '../modules/local/mutserve'
 include { ANNOTATE } from '../modules/local/annotate'
 include { HAPLOGROUP_DETECTION } from '../modules/local/haplogroup_detection'
@@ -29,6 +30,14 @@ workflow MITOCALLING {
         exit 1, "Reference " + params.reference + "not supported"
     }
 
+    CALCULATE_STATISTICS(
+        bams_ch
+    )
+
+    SUMMARIZE_STATISTICS(
+        CALCULATE_STATISTICS.out.stats_ch.collect()
+    )
+
     MUTSERVE(
         bams_ch.collect(),
         ref_file