Hi all, When running the following command for a mouse tumor-only sample using nf-core/sarek: nextflow run nf-core/sarek -r 3.5.1 \ -profile singularity \ --input SAMPLE_SHEET \ --outdir somatic \ --step mapping \ --save_mapped \ --genome GRCm38 \ --tools mutect2,cnvkit,vep \ --vep_cache ./vep_cache \ --vep_species mus_musculus -resume I downloaded the file https://ftp.ensembl.org/pub/release-102/variation/indexed_vep_cache/mus_musculus_vep_102_GRCm38.tar.gz and extracted it into the

./vep_cache

directory. After running the pipeline, the

somatic

outdir contains subfolders such as

csv

,

multiqc

, and

variant_calling

, but no

annotation

subfolder. The pipeline completed successfully without any errors. However, when using the parameter "--tools haplotypecaller,vep", the output directory does include an

annotation

subfolder. How to solve the issue? Thanks!

I am pretty sure this is because of a bug that has been around for a while and is specific to mouse and other organisms that lack a

germline_resource

file. (If you want, you can confirm that by adding

stelka

and/or

freebayes

to your tools. The bug affects

mutect2

, and your only other

tool

is

cnvkit

, the output of which cannot be meaningfully annotated with VEP.) Basically what happens is that the built-in filtering for Mutect2 uses

germline_variant

and sarek does not accommodate the case where that is not available (even though you could run the actual filtering command from GATK without it, at the cost of not filtering everything that you would get with the resource). This results in the Mutect2 VCF files lacking a populated

FILTER

column, and only the

*.mutect2.vcf.gz

files are generated, while the

*.mutect2.filtered.vcf.gz

files are missing in the output. Those latter files, however, are the input for the

annotation

--step

, i.e. for VEP, so without them it will not run. This is how I have worked around the problem so far (if I find time, I plan to actually fix it): • Run the pipeline to get the (unfiltered) Mutect2 files. (You have already done that.) • Filter the Mutect2 files "manually" (see below) to generate the input files for the annotation. • Rerun the pipeline starting from

--step annotate

with

--tools vep

and an

--input

CSV that points to the filtered VCFs you just created. Run something like this inside the GATK container from the pipeline to do the filtering (for each file):

fasta=/path/to/iGenomes/Mus_musculus/Ensembl/GRCm38/Sequence/WholeGenomeFasta/genome.fa
vcf=/path/to/somesample.mutect2.vcf.gz # from 1st sarek run
filtvcf="${vcf%vcf.gz}filtered.vcf.gz"

gatk --java-options "-Xmx9830M -XX:-UsePerfData" \
    FilterMutectCalls \
    --variant "$vcf" \
    --output "$filtvcf" \
    --reference "$fasta" \
    --orientation-bias-artifact-priors "${vcf%vcf.gz}artifactprior.tar.gz" \
    --tmp-dir .

Hope this makes sense, let me know if you have questions.

@Thomas Danhorn Thank you very much for your guidance. After creating a new samplesheet with a “vcf” column instead of the previous “fastq_1” and “fastq_2” columns, and running the Sarek pipeline with

--step annotate

, the output now adds the

annotation

subfolder. I truly appreciate your time and support!