Merge pull request #14 from WEHI-ResearchComputing/flagstat

Flagstat

config/modules.config

@@ -54,6 +54,11 @@ process {
         memory = { 4.GB * task.attempt }
         time = '2h'
     }
+    withLabel: Flagstats {
+        cpus = 1
+        memory = { 4.GB * task.attempt }
+        time = '1h'
+    }
     withLabel: Multiqc {
         cpus = 2
         memory = { 1.GB * task.attempt }

config/multiqc_config.yaml

@@ -5,6 +5,7 @@ remove_sections:
   - fastqc_per_sequence_gc_content
   - fastqc_adapter_content
   - fastqc_per_base_sequence_content
+  - fastqc_sequence_length_distribution
   - mosdepth-coverage-dist-cov
 mosdepth_config:
   general_stats_coverage:
@@ -13,3 +14,5 @@ mosdepth_config:
   exclude_contigs:
     - "*_API_*"
     -  "*_M*"
+table_columns_visible:
+    mapped_passed_pct: true

main.nf

@@ -29,6 +29,7 @@ include { RPlotROI } from './modules/stvariant.nf'
 
 include{ FastQC } from './modules/qc.nf'
 include{ MosDepth } from './modules/qc.nf'
+include{ FlagStats } from './modules/qc.nf'
 include{ MultiQC } from './modules/qc.nf'
 
 process foo {
@@ -134,8 +135,9 @@ workflow {
     fastqc_ch=FastQC(bam_ch.bamnodup.map{row->row[1]}.unique({it.baseName}).collect())
 
     mosdepth_ch=MosDepth(bam_ch.bamnodup.join(bam_ch.bai)) // join bams with their index based on groupId
+    flagstat_ch=FlagStats(bam_ch[0], bam_ch.bamnodup.map{row->row[1]})
 
-    MultiQC( fastqc_ch.zip.mix(mosdepth_ch).collect().ifEmpty([]) )  
+    MultiQC( fastqc_ch.zip.mix(mosdepth_ch, flagstat_ch).collect().ifEmpty([]) )  
     //----------------BCF tools----------------------------------------
     //BCF Input Channel Emits parentId,groupId,ref,bamlist,bams,parentbams
     input_ch.map{row -> tuple(row[0],row[4],row[5])}

modules/qc.nf

@@ -1,6 +1,6 @@
 process FastQC {
     label 'Fastqc' // assumes resources are defined in a label
-    tag "${groupId}"
+    tag "${sampleId}"
     publishDir "${params.outdir}/QC", mode: 'copy'
     cache true
     input:
@@ -19,7 +19,7 @@ process FastQC {
 
 process MosDepth {
     label 'Mosdepth'
-    tag "${groupId}"
+    tag "${sampleId}"
     publishDir "${params.outdir}/QC", mode: 'copy'
     cache true
     input:
@@ -35,6 +35,25 @@ process MosDepth {
     """
 }
 
+process FlagStats {
+    label 'Flagstats'
+    tag "${sampleId}"
+    publishDir "${params.outdir}/QC", mode: 'copy'
+    cache true
+    input:
+    val(sampleId)
+    path(bam)
+
+    output:
+    path("*.flagstats"), emit: txt
+
+
+    script:
+    """
+    samtools flagstats ${bam} > ${bam.baseName}.flagstats
+    """
+}
+
 process MultiQC {
     label 'Multiqc'
     cache true

nextflow.config

@@ -35,7 +35,7 @@ profiles {
         docker.enabled          = false
         singularity.pullTimeout = "3h"
         process {
-            withLabel: 'BWAAlign|Samtools|Bcftools|Fastqc|Mosdepth|Multiqc|RGridss|Rfilter|IndexDedup' {
+            withLabel: 'BWAAlign|Samtools|Bcftools|Fastqc|Mosdepth|Flagstats|Multiqc|RGridss|Rfilter|IndexDedup' {
                 container = "ghcr.io/wehi-researchcomputing/nf-malaria-variant-analysis:latest"
             }
         }
@@ -45,7 +45,7 @@ profiles {
         apptainer.enabled     = false
         singularity.enabled   = false
         process {
-            withLabel: 'BWAAlign|Samtools|Bcftools|Fastqc|Mosdepth|Multiqc|RGridss|Rfilter|IndexDedup' {
+            withLabel: 'BWAAlign|Samtools|Bcftools|Fastqc|Mosdepth|Flagstats|Multiqc|RGridss|Rfilter|IndexDedup' {
                 container = "ghcr.io/wehi-researchcomputing/nf-malaria-variant-analysis:latest"
             }
         }