[ensembl-dev] Choosing the --per_gene option in VEP

Aravind Sankar as42 at sanger.ac.uk
Thu Oct 20 13:35:01 BST 2016


Hello,

I have been using the --per_gene option in VEP to get only the most severe consequence of each gene in the CSQ field for my files (as given in the documentation here http://www.ensembl.org/info/docs/tools/vep/script/vep_options.html#opt_per_gene). However, I noticed something going awry today. I was checking all the consequences of a particular position in a gene (BAP1). When I give the --per_gene option in vep, it returns a single consequence for the variant and the consequence is an intron variant. When I don’t give the –per_gene option, it returns all the possible consequences for BAP1, which includes both a missense variant and an intron variant. For some reason, the missense variant is not being reported over the intron variant while using the –per_gene option. I ran this on build 84. The commands I used were as follows:-

Using per_gene :

perl /software/vertres/bin-external/variant_effect_predictor_v84.pl --offline --vcf --dir_cache /lustre/scratch116/vr/ref/ensembl/vep_cache --species homo_sapiens --assembly GRCh38 --no_progress --everything --cache --per_gene -i bap1_pos_test.vcf -o bap1_84_per_gene.vcf

Without per_gene :

perl /software/vertres/bin-external/variant_effect_predictor_v84.pl --offline --vcf --dir_cache /lustre/scratch116/vr/ref/ensembl/vep_cache --species homo_sapiens --assembly GRCh38 --no_progress --everything --cache -i bap1_pos_test.vcf -o bap1_84.vcf

The output of the per_gene file doesn’t have the missense variant while the normal one does, for the same gene.
less bap1_84_per_gene.vcf | grep -c "missense"
0

less bap1_84.vcf | grep -c "missense"
1

This is contrary to what is expected from the given documentation. Could someone help clarify what is going on here ?

Thanking you,
Aravind




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ensembl.org/pipermail/dev_ensembl.org/attachments/20161020/21237116/attachment.html>


More information about the Dev mailing list