Ensembl VEP is skipped when using custom a custom genome (--igenomes_ignore --genome null)

[amizeranschi]

Description of the bug

Using Nextflow v. 23.04.1 and sarek v. 3.2.0, Ensembl VEP is being executed fine with a built-in genome, but it ends up getting silently skipped (with no error or warning in the logs) with a custom genome for the same species. I tested this in yeast, as follows:

With a custom genome, the command below skips VEP:

nextflow run nf-core/sarek -r 3.2.0 \
--skip_tools haplotypecaller_filter,baserecalibrator,markduplicates \
-profile docker \
--input sample-sheet.csv \
--outdir sarek-test-jointVC \
--max_time '336.h' \
--max_cpus '64' \
--max_memory '256.GB' \
--fasta ${main_dir}/reproducible-sarek-error/Saccharomyces_cerevisiae.R64-1-1-chr.fa \
--fasta_fai ${main_dir}/reproducible-sarek-error/Saccharomyces_cerevisiae.R64-1-1-chr.fa.fai \
--dict ${main_dir}/reproducible-sarek-error/Saccharomyces_cerevisiae.R64-1-1-chr.dict \
--save_mapped \
--vep_cache ${main_dir}/reproducible-sarek-error/Saccharomyces_cerevisiae \
--bwa ${main_dir}/reproducible-sarek-error/bwa-index \
--igenomes_ignore \
--genome null \
--tools haplotypecaller,vep \
--vep_out_format tab \
--vep_include_fasta

With the built-in genome, VEP is being executed:

nextflow run nf-core/sarek -r 3.2.0 \
--skip_tools haplotypecaller_filter,baserecalibrator,markduplicates \
-profile docker \
--input sample-sheet.csv \
--outdir sarek-test-jointVC \
--max_time '336.h' \
--max_cpus '64' \
--max_memory '256.GB' \
--save_mapped \
--genome R64-1-1 \
--tools haplotypecaller,vep \
--vep_out_format tab \
--vep_include_fasta

I am including two log files below, one for a test in which VEP was skipped and one where VEP was executed.

Command used and terminal output

No response

Relevant files

this-skipped-vep.log.txt

this-ran-vep.log.txt

System information

No response

[Adrian-Zet]

Hello amizeranschi,

I am writing this pull request to get some feedback regarding the amizeranschi/bcbio-full (https://hub.docker.com/r/amizeranschi/bcbio-full/tags)
We know that bcbio-vm docker image has been discontinued.
Thus in the absence of any documentation we want to know if amizeranschi/bcbio-full docker image is a functional image and can be installed to run bcbio in a containerized version. Also, in this case, how are the databases installed, must they be installed outside of the docker image?

Many thanks,
Adrian

[amizeranschi]

As it turns out, this was no bug, but just a configuration error. The custom genome setup required some additional VEP parameters (hidden by default in the documentation) which were set up automatically for the built-in genome.

For example: https://nf-co.re/sarek/3.2.1/parameters#vep_genome.