VEP order most severe to least

Question

[shwethacm]

Hi there,

I have a rather large VCF file annotated using VEP, and I am trying to pick the most severe consequence of the variant for each variant in the file. Now I know that VEP automatically arranges the output to be listed as most-severe to least according to this (http://uswest.ensembl.org/info/genome/variation/prediction/predicted_data.html#consequences). However, my VCF doesn't have it in that format (no idea why, but I spot checked, and it doesn't) .

I cannot re-make the VCF using --severe because I didn't create it in the first place. My solution is to write a script to do this manually. But before I do that, I'd be interested if anyone out there has encountered a similar situation and have better suggestions for this.

Thanks for the help!

Answer 1

[Ben_Ensembl]

Hi shwethacm,

I'm afraid that I'm not able to help with any particular suggestions or custom scripts that will help to filter or rearrange your VCF output. However, I wanted to correct a small error in your assumption- the VEP does not necessarily arrange the output from most to least severe for each variant.

Some considerations for your filtering and reordering:

• You may be able to run your VCF through the VEP again using the different --pick options: http://www.ensembl.org/info/docs/tools/vep/script/vep_options.html#filt
• You may be able to use the filter_vep script to filter your results: http://www.ensembl.org/info/docs/tools/vep/script/vep_filter.html

Best wishes

Ben Ensembl Helpdesk

[shwethacm]

Thanks Ben!! --pick did what I needed. I ended up doing this:

1. Filter out variants from my region of interest

2. Run VEP again like so:

   vep -i {input} -o {output} --cache --merged --pick --pick_order biotype,ccds,rank,canonical --force_overwrite --vcf

My aim is to filter out variants protein coding regions, and most damaging variants . The default --pick order missed a few because of overlapping transcripts, so I ended up reordering the 'pick' criteria to biotype,ccds,rank,canonical . This worked great! Thanks you for the prompt answer!