Update pipeline to run with a small test dataset

CHANGELOG.md

@@ -10,4 +10,5 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 ### Added
 
+- Update pipeline to run with a small test dataset [#35](https://github.com/genomic-medicine-sweden/skierfe/pull/35)
 - Added test data and test profile [#33](https://github.com/genomic-medicine-sweden/skierfe/pull/33)

conf/modules/assembly_variant_calling.config

@@ -13,6 +13,9 @@
 process {
 
     withName: DIPCALL {
+
+        ext.args = '--cs'
+
         publishDir = [
             path: { "${params.outdir}/assembly_variant_calling/dipcall/${meta.id}" },
             mode: params.publish_dir_mode,

conf/test.config

@@ -49,6 +49,10 @@ params {
 }
 
 process {
+    withName: '.*:ASSEMBLY:HIFIASM' {
+
+        ext.args = '-f0'
+    }
     withName: '.*:PHASING:WHATSHAP_PHASE' {
 
         ext.args = '--ignore-read-groups --indels --distrust-genotypes --include-homozygous'

modules.json

@@ -60,7 +60,7 @@
                     },
                     "fastqc": {
                         "branch": "master",
-                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
+                        "git_sha": "f4ae1d942bd50c5c0b9bd2de1393ce38315ba57c",
                         "installed_by": ["modules"],
                         "patch": "modules/nf-core/fastqc/fastqc.diff"
                     },
@@ -75,6 +75,11 @@
                         "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
                         "installed_by": ["modules"]
                     },
+                    "gunzip": {
+                        "branch": "master",
+                        "git_sha": "3a5fef109d113b4997c9822198664ca5f2716208",
+                        "installed_by": ["modules"]
+                    },
                     "hifiasm": {
                         "branch": "master",
                         "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
@@ -109,11 +114,6 @@
                         "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
                         "installed_by": ["modules"]
                     },
-                    "samtools/fastq": {
-                        "branch": "master",
-                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
-                        "installed_by": ["modules"]
-                    },
                     "samtools/index": {
                         "branch": "master",
                         "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",

modules/local/dipcall.nf

@@ -41,17 +41,17 @@ process DIPCALL {
     def prefix = task.ext.prefix ?: "${meta.id}"
     if(sex == '2') {
     """
-    minimap2 -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.paf.gz.log | gzip > ${prefix}.hap1.paf.gz
-    minimap2 -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.paf.gz.log | gzip > ${prefix}.hap2.paf.gz
+    minimap2 ${args} -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.paf.gz.log | gzip > ${prefix}.hap1.paf.gz
+    minimap2 ${args} -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.paf.gz.log | gzip > ${prefix}.hap2.paf.gz
 
-    minimap2 -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.sam.gz.log | gzip > ${prefix}.hap1.sam.gz
-    minimap2 -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.sam.gz.log | gzip > ${prefix}.hap2.sam.gz
+    minimap2 ${args} -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.sam.gz.log | gzip > ${prefix}.hap1.sam.gz
+    minimap2 ${args} -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.sam.gz.log | gzip > ${prefix}.hap2.sam.gz
 
     gzip -dc ${prefix}.hap1.paf.gz | sort -k6,6 -k8,8n | k8 /usr/local/bin/paftools.js call - 2> ${prefix}.hap1.var.gz.vst | gzip > ${prefix}.hap1.var.gz
     gzip -dc ${prefix}.hap2.paf.gz | sort -k6,6 -k8,8n | k8 /usr/local/bin/paftools.js call - 2> ${prefix}.hap2.var.gz.vst | gzip > ${prefix}.hap2.var.gz
 
-    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap1.sam.gz | samtools sort -m4G --threads ${task.cpus/2} -o ${prefix}.hap1.bam -
-    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap2.sam.gz | samtools sort -m4G --threads ${task.cpus/2} -o ${prefix}.hap2.bam -
+    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap1.sam.gz | samtools sort --threads ${task.cpus} -o ${prefix}.hap1.bam -
+    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap2.sam.gz | samtools sort --threads ${task.cpus} -o ${prefix}.hap2.bam -
 
     gzip -dc ${prefix}.hap1.var.gz | grep ^R | cut -f2- > ${prefix}.hap1.bed
     gzip -dc ${prefix}.hap2.var.gz | grep ^R | cut -f2- > ${prefix}.hap2.bed
@@ -69,17 +69,17 @@ process DIPCALL {
     """
     } else if (sex == '1') {
     """
-    minimap2 -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.paf.gz.log | gzip > ${prefix}.hap1.paf.gz
-    minimap2 -c --paf-no-hit -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.paf.gz.log | gzip > ${prefix}.hap2.paf.gz
+    minimap2 ${args} -c --paf-no-hit -xasm5  -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.paf.gz.log | gzip > ${prefix}.hap1.paf.gz
+    minimap2 ${args} -c --paf-no-hit -xasm5  -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.paf.gz.log | gzip > ${prefix}.hap2.paf.gz
 
-    minimap2 -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.sam.gz.log | gzip > ${prefix}.hap1.sam.gz
-    minimap2 -a -xasm5 --cs -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.sam.gz.log | gzip > ${prefix}.hap2.sam.gz
+    minimap2 ${args} -a -xasm5  -t ${task.cpus} ${mmi} ${haplotype_1} 2> ${prefix}.hap1.sam.gz.log | gzip > ${prefix}.hap1.sam.gz
+    minimap2 ${args} -a -xasm5  -t ${task.cpus} ${mmi} ${haplotype_2} 2> ${prefix}.hap2.sam.gz.log | gzip > ${prefix}.hap2.sam.gz
 
     gzip -dc ${prefix}.hap1.paf.gz | sort -k6,6 -k8,8n | k8 /usr/local/bin/paftools.js call - 2> ${prefix}.hap1.var.gz.vst | gzip > ${prefix}.hap1.var.gz
     gzip -dc ${prefix}.hap2.paf.gz | sort -k6,6 -k8,8n | k8 /usr/local/bin/paftools.js call - 2> ${prefix}.hap2.var.gz.vst | gzip > ${prefix}.hap2.var.gz
 
-    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap1.sam.gz | samtools sort -m4G --threads ${task.cpus} -o ${prefix}.hap1.bam -
-    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap2.sam.gz | samtools sort -m4G --threads ${task.cpus} -o ${prefix}.hap2.bam -
+    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap1.sam.gz | samtools sort --threads ${task.cpus} -o ${prefix}.hap1.bam -
+    k8 /usr/local/bin/dipcall-aux.js samflt ${prefix}.hap2.sam.gz | samtools sort --threads ${task.cpus} -o ${prefix}.hap2.bam -
 
     gzip -dc ${prefix}.hap1.var.gz | grep ^R | cut -f2- > ${prefix}.hap1.bed
     gzip -dc ${prefix}.hap2.var.gz | grep ^R | cut -f2- > ${prefix}.hap2.bed

modules/local/modkit/pileup/main.nf

@@ -2,12 +2,13 @@ process MODKIT_PILEUP {
     tag "$meta.id"
     label 'process_high'
 
-    container "docker.io/fellen31/modkit:0.1.9"
+    container "docker.io/fellen31/modkit:0.2.5"
 
     input:
     tuple val(meta), path(bam), path(bai)
     tuple val(meta2), path(fasta)
     tuple val(meta3), path(fai)
+    tuple val(meta4), path(bed)
 
     output:
     tuple val(meta), path("${meta.id}.bed"), emit: bed, optional: true
@@ -24,19 +25,22 @@ process MODKIT_PILEUP {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+
+    def include_bed = bed ? "--include-bed ${bed}" : ''
     """
     modkit pileup \\
         $args \\
+        $include_bed \\
         --threads ${task.cpus} \\
         --ref $fasta \\
         --log-filepath ${bam.baseName}.modkit.log \\
         ${bam} \\
         ${meta.id}.${bam.baseName}.bed
 
     if test -d ${meta.id}.${bam.baseName}.bed; then
-        mv ${meta.id}.${bam.baseName}.bed/1.bed ${meta.id}.1.bed
-        mv ${meta.id}.${bam.baseName}.bed/2.bed ${meta.id}.2.bed
-        mv ${meta.id}.${bam.baseName}.bed/ungrouped.bed ${meta.id}.ungrouped.bed
+        if test -f ${meta.id}.${bam.baseName}.bed/1.bed; then mv ${meta.id}.${bam.baseName}.bed/1.bed ${meta.id}.1.bed; fi
+        if test -f ${meta.id}.${bam.baseName}.bed/2.bed; then mv ${meta.id}.${bam.baseName}.bed/2.bed ${meta.id}.2.bed; fi
+        if test -f ${meta.id}.${bam.baseName}.bed/ungrouped.bed; then mv ${meta.id}.${bam.baseName}.bed/ungrouped.bed ${meta.id}.ungrouped.bed; fi
     else
         mv ${meta.id}.${bam.baseName}.bed ${meta.id}.bed
     fi