bccdc-phl / call-variants

Align against a reference genome and call variants

Call Variants Pipeline

This pipeline takes a directory containing paired-end illumina sequence data, plus a reference sequence, aligns the reads against the reference and calls variants. It is based directly off of the excellent Snippy by Torsten Seemann (@tseemann).

Usage

nextflow run BCCDC-PHL/call-variants \
  --ref <ref.fa> \
  --gff <genes.gff> \
  --fastq_input </path/to/fastq> \
  --outdir <outdir>

call-variants's Issues

Task CPUs missing

the cpus directive was not included in this process:

call-variants/modules/call_variants.nf

Lines 73 to 78 in 17df90f

	bwa mem -t ${task.cpus} -Y -M -R "${rg_string}" ${ref_data}/ref/sequences.fa ${reads} | \
	samclip --max ${params.maxsoft} --ref ${ref_data}/ref/sequences.fa.fai | \
	samtools sort -@ ${task.cpus} -n -l 0 | \
	samtools fixmate -@ ${task.cpus} -m - - | \
	samtools sort -@ ${task.cpus} -l 0 | \
	samtools markdup -@ ${task.cpus} -r - - \

Also, be careful that when you chain processes like this is that they stream to each other so they will each be running in parallel with X CPUs, so you should make sure that if you ask for X, that you aren't using N times X because of all the processes. An easy way is to split the CPUs requested equally to each of the processes in the chain.

Moving away from VT

We are moving away from VT to bcftools norm:

call-variants/modules/call_variants.nf

Line 112 in 17df90f

vt normalize -r ${ref_data}/ref/sequences.fa - | \

Moving away from snpEff

We are looking into bcftools csq instea:

call-variants/modules/call_variants.nf

Lines 138 to 155 in 17df90f

	process annotate_variants {
	tag { sample_id }
	publishDir "${params.outdir}/${sample_id}", mode: 'copy', pattern: "${sample_id}.vcf"
	input:
	tuple val(sample_id), path(filtered_vcf), val(ref_id), path(ref_data), path(snpeff_config)
	output:
	tuple val(sample_id), path("${sample_id}.vcf")
	script:
	"""
	snpEff ann -noLog -noStats -no-downstream -no-upstream -no-utr \
	-c ${snpeff_config} -dataDir ${ref_data} ref ${filtered_vcf} > ${sample_id}.vcf
	"""
	}