Variants missing from VEP cache

[suzyhh]

Hello,

I've been running VEP v113 using the cache and docker image, and I've found some variants that are present in gnomAD v4.1 but are absent from the cache.

Example 1: 20:45891598:CCTG:C the co-located variant is flagged in Ensembl, I have now added in --failed 1 to my VEP command and I expect this variant will now have AFs.

Example 2: 8:144392334:T:TGGGGGTGCAAGGTGA this variant is present in gnomAD but is not in dbSNP

Example 3: 1:9244919:CCCCAGGCA:C this variant is present in gnomAD and ClinVar, but is not in dbSNP

On this page: https://useast.ensembl.org/info/docs/tools/vep/script/vep_cache.html#cache in the "Data in the cache" section, it lists the sources of variants in the VEP cache, and this includes dbSNP, ClinVar, and gnomAD v4.1. I would therefore assume that any variant in any of these sources should be present in the cache, however what I've observed is that the variant has to be present in dbSNP to be in the cache - which is the expected behaviour?

System

• VEP version: 113.2 docker
• VEP Cache version: 113

Full VEP command line

        vep --cache --offline \
        -i ~{vcf} \
        --dir_cache /opt/vep/.vep \
        --dir_plugins /opt/vep/.vep/Plugins/ \
        --vcf --compress_output bgzip \
        --merged \
        --fasta $refFasta \
        --assembly GRCh38 \
        --no_stats \
        --fork ~{fork} \
        --buffer_size ~{buffer} \
        --no_escape \
        --check_existing --failed 1 \
        --hgvs --hgvsg \
        --af --af_gnomadg \
        --protein --uniprot \
        --symbol \
        --numbers \
        --allele_number \
        --sift b --polyphen b \
        --pubmed \
        --show_ref_allele \
        --variant_class \
        --transcript_version \
        --mane \
        --flag_pick_allele_gene --pick_order rank,mane_select,mane_plus_clinical,canonical,appris,tsl,biotype,ccds,length \
        --flag_gencode_primary \
        --exclude_predicted \
        --use_given_ref \
        --plugin MaxEntScan,/opt/vep/.vep/fordownload,NCSS,SWA \
        --plugin SpliceAI,snv=~{spliceaiSnv},indel=~{spliceaiIndel} \
        --plugin SpliceRegion \
        --plugin NearestExonJB,max_range=100 \
        --plugin REVEL,file=/opt/vep/.vep/new_tabbed_revel_grch38.tsv.gz,no_match=1 \
        --plugin SpliceDistance \
        --plugin UTRAnnotator,file=/opt/vep/.vep/fordownload/uORF_5UTR_GRCh38_PUBLIC.txt \
        --plugin GnomadPli,file=~{gnomadv4Pli} \
        --plugin CADD,snv=~{caddSnv},indels=~{caddIndel} \
        --plugin AlphaMissense,file=~{alphaMissense} \
        --custom file=${hgmdZip},short_name=HGMD,format=vcf,type=exact,coords=0,fields=CLASS%PHEN \
        --custom file=~{clinVarVcf},short_name=ClinVar,format=vcf,type=exact,coords=0,fields=CLNSIG%CLNREVSTAT%CLNDN \
        --force_overwrite \
        -o ~{outName}_snvs.vep.vcf.gz

I would really like to avoid having to add the gnomAD AFs as a custom annotation as I imagine this is going to add quite a bit of processing time to the annotation, but we're missing AFs from quite a few variants which is having a knock-on effect for our variant filtering processes.

Thanks!
Suzy

[dglemos]

Hi @suzyhh,
That is correct, variants must be present in dbSNP to be included in the cache. For VEP cache version 113 the corresponding dbSNP version is 156.
Unfortunately, the only way to obtain the gnomAD annotation is by using the custom annotation.

Testing the example variants with custom annotation, I can find the following matches:

• 8:144392334:T:TGGGGGTGCAAGGTGA (gnomAD link)
• 20:45891598:CCTG:C (gnomAD link)

However, I cannot find gnomAD data for 1:9244919:CCCCAGGCA:C. Can you please send the expected data ?