Annotations vep : keep high impact in any transcript + canonical

Question

[quentin54520]

Hello all,

I use vep to annotate my vcf. I would like to know if there is a way, if a variants as a high impact in a non canonical transcript, to keep it and the annotations in the canonical. I don't want to annotate any impact in any transcript to avoid huge amount of data, but i don't want to miss a high impact in alternative transcript.

[Ram]

Please note:

1. It's spelled canonical (root word "canon"), not cannon-ical. There's no such word.
2. It's spelled transcript with a 'p' before the 't'.

I've fixed these in your post and title.

Answer 1

[Ben_Ensembl]

Hi quentin54520,

I would advise annotating your variants with the VEP without any filters, then using the filter_vep tool to filter your VEP output: https://www.ensembl.org/info/docs/tools/vep/script/vep_filter.html#filter_run

You can write custom filters to return lines that refer to either the canonical transcript or have a particular IMPACT rating.

[Ram]

Side note: Canonical isn't always what people may be looking for. For example, the longest transcript for ESR1 and BRCA1 genes are not the most popular ones. Canonical is mostly but not always what we are looking for. Canonical translates to "longest", not "isoform 1 on NCBI". For that, you may want to use the MANE + RefSeq Select.

[Ben_Ensembl]

This is very true, Ram. Here is the Ensembl documentation on canonical transcript selection: https://www.ensembl.org/info/genome/genebuild/canonical.html

[Ram]

Thank you, Ben. Another super-useful feature is being able to assign priority to flag-pick-allele. Use the option to flag an entry based on one's own priorities: https://useast.ensembl.org/info/docs/tools/vep/script/vep_options.html#opt_pick_order

The option to flag something helps us get VEP's take but ask VEP to not throw anything away on its own discretion.

[quentin54520]

Yes, indeed it is very difficult to know what a canonical transcript is, there can be several definitions depending on the people and the context. I work on rare diseases, and I have to annotate human genomes. Usually I use the --pick option but I'm always afraid to miss a potentially interesting variant. But on the other hand I don't want to end up with ten annotations per variant, there are already so many variants. Do you think there would be a better solution than the --pick option? Using the --mane option and then filtering to keep only what has the mane flag?

[Ram]

Use --flag options instead of plain --pick and get as much annotation info as you can. Perform filtering later once you have determined the optimal filters to remove noise while retaining important information. That's the most advice we can give you without getting into the particulars of your requirements.