Ensembl vep GRCh38 cache version

Question

[sg197]

Hi,

I've performed variant calling using the GATK GRCh38 genome reference. The reference is gencode v27 (ensembl version 90 from gtf header), and my understanding is that when using ensembl vep, you should use the same cache version as was used for alignment. I downloaded GRCh38 v90 following instructions from ensembl vep to get the correct cache

curl -O https://ftp.ensembl.org/pub/release-90/variation/indexed_vep_cache/homo_sapiens_vep_90_GRCh38.tar.gz tar xzf homo_sapiens_vep_90_GRCh38.tar.gz

(The expanded tar then contains homo_sapiens/90_GRCh38)

I then run ensembl vep as such:

vep -i $inputvcf --dir_cache /path/to/90_GRCh38 --cache --cache_version 90 -o $outputvcf

However on the html output from vep it says the annotation source as "Cache: /home/Software/ensembl-vep/Databases/GRCh38_V90_GATK/homo_sapiens/90_GRCh38; homo_sapiens_core_111_38 on ensembldb.ensembl.org"

My VEP version is 111, and I'm having issues re-downloading an older version of ensembl-vep. Is it possible to run backwards compatible caches with a newer version of ensembl vep? It seems to be using v111 from ensembldb.enembl.org despite me setting the version and cache dir.

Any help appreciated, thanks!

[GenoMax]

Tagging: Louisse_Ensembl

Answer 1

[Jeremy Leipzig]

my understanding is that when using ensembl vep, you should use the same cache version as was used for alignment

where does it say this? what does annotation have to do with alignment, or variant calling for that matter?

The vcfs are based on GRCh38. That's the freeze. That's it. The alignment and variant calling steps don't know what genes are.

Maybe you mean matching VEP software version to a specific version of the annotation sources?

[sg197]

Ahh yes! I had misread the matching cache and annotation sources statement. So it is ok to run VEP v111 with the v111 cache, against vcf's generated from bam's aligned to GATK GRCh38. Thank you for clarifying!