Sarek did not perform variant calling?

Question

[Joe]

I'm trying to check for mutations from whole exome sequencing of two samples from the same patient, and was recommended to use the nextflow sarek pipeline. I assembled the fastq files I needed, made the csv file describing the patient sample information (patient, sample, lane, fastq_1, fastq_2), and entered the following command:

nextflow run nf-core/sarek
-profile docker
--input ./samplesheet.csv
--outdir .

This ran for 12 hours and gave me the message

- [nf-core/sarek] Pipeline completed successfully-

The output files/folders have appeared as described in the "Directory Structure" described here: https://nf-co.re/sarek/3.0.1/docs/output

In the workflow report in the pipeline_info folder I see the following processes were run:

• FASTQC
• TABIX_BGZIPTABIX_INTERVAL_COMBINED
• CREATE_INTERVALS_BED
• TABIX_BGZIPTABIX_INTERVAL_SPLIT
• FASTP
• BWAMEM1_MEM
• GATK4_MARKDUPLICATES
• MOSDEPTH
• SAMTOOLS_STATS
• GATK4_BASERECALIBRATOR
• GATK4_GATHERBQSRREPORTS
• GTK4_APPLYBQSR
• MERGE_CRAM
• INDEX_CRAM
• CUSTOM_DUMPSOFTWAREVERSIONS
• MULTIQC

None of these are for variant calling as I understand. The output page (https://nf-co.re/sarek/3.0.1/docs/output), under "Variant Calling" says that results regarding variant calling are collected in {outdir}/variantcalling/, however this folder does not exist for me. So the pipeline has completed, but has not performed any variant calling. Why has variant calling not been performed and how do I perform variant calling with sarek here?

Thanks in advance

Answer 1

[Maxime Garcia]

It's by design, we only do variant calling on the tools specified with the tools params, cf https://nf-co.re/sarek/3.2.3/parameters#tools

[ATpoint]

The usage docs say that the standard command for quickstarters is

nextflow run nf-core/sarek --input samplesheet.csv --outdir <OUTDIR> --genome GATK.GRCh38 -profile docker

which OP is using. Does that mean that the quickstart command for a variant calling pipeline does not call any variants? If so docs should emphasize this more.

[Maxime Garcia]

I think it's clear enough from the docs: https://nf-co.re/sarek/3.2.3#pipeline-summary

As we allow multiple variant calling tools, that might or might not be useful depending on the use case, we decided not to specify any by default.

Adding an extra warning on the command line when no tools are specified could be an idea, I'll add that as an issue.

We also thought about having profile that are preconfigured for certain analysis.

I'll speak with the other devs and we'll look into it.

[ATpoint]

I think it's clear enough from the docs

I admit that in Introduction it is stated clearly. My suggestion is to add this clearly to the top of the Usage section as well though.

Made a PR, https://github.com/nf-core/sarek/pull/1166

[Maxime Garcia]

Thanks for the PR <3 Welcome to the team \o/