This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Protein coding transcripts with vep

I'm using VEP to annotate variants in VCF files. In the analysis of the results I want to select only the protein coding transcripts. Should I look at the field "consequence" added by VEP? I don't have a biological background and I didn't find any answer when searching on the internet until now.

vep annotation variants
• 5,508 views • link
updated by abio • abio • written by onemoreuser • onemoreuser

In my output VEP txt file, I don't see a "protein-coding" category. I only see the following:

table(d$BIOTYPE)

-      CTCF_binding_site enhancer  open_chromatin_region    promoter   promoter_flanking_region  TFBS 
10527        109                       39                 42                       103              167                       34

However if I look at which of these variants are found in CCDS or APPRIS, I find the following.Kindly advise on how to then filter out protein coding variants?

table(d$BIOTYPE,d$APPRIS != "-")

                         -   A1   A2   P1   P2   P3   P4   P5
 -                        7903  150  760 1123  156  337   75   23
 CTCF_binding_site         109    0    0    0    0    0    0    0
enhancer                   39    0    0    0    0    0    0    0
open_chromatin_region      42    0    0    0    0    0    0    0
promoter                  103    0    0    0    0    0    0    0
promoter_flanking_region  167    0    0    0    0    0    0    0
TF_binding_site            34    0    0    0    0    0    0    0
• 0 views • link
written by abio • abio

2 answers

I usually run VEP with option "--everything". In the results you can see the consequence for each overlapping transcript. You have to carefully parse the results, which is a bit complicate when there are multiallelic variants. There is a field for transcript-biotype named "BIOTYPE" and another field for the transcript ("Feature"). Just set a filter for BIOTYPE to be "protein_coding".

Alternatively, you may preload a list of protein-coding transcripts (you can get them from Biomart), and see whether the transcript in the "Feature" field is within your list. BTW, I would also recommend not to use all the protein-coding transcripts but only the ones that are more reliable (e.g. with CCDS or APPRIS support).

[edited to correct a detail]

• 1 views • link
written by abascalfederico abascalfederico

Thank you. One of the requirements is to be a protein coding transcript but I also filter other features like you said.

• 1 views • link
written by onemoreuser • onemoreuser

You can filter your output by consequence type. Are you using the online tool or the standalone script? On the online tool, there's a little filter box above your results table where you can select Biotype is protein_coding. If you're using the script, you can run the filter script with --filter "Biotype is protein_coding".

• 1 views • link
written by Emily Emily

Thanks Emily, as someone who is using the script this is exactly what I needed.

• 0 views • link
written by brismiller • brismiller

Log in to answer this question.

More posts like this