Representative AA change from point mutation given by genomic lcus

Question

[sak042]

We are having a hard time to determine a "representative" amino acid change that resulted from a genomic point mutation. Obviously, a genomic point mutation in a coding region leads to a protein sequence change (including missense, nonsense, splice-variants... etc). Problem is, there are multiple possible transcripts (from a same gene) that are affected from the point mutation, and the result can be different for each transcript. For example, one mutation can be synonymous in transcript A, but missense in transcript B and nonsense in transcript C. Question is, if we need to pick one AA change, which transcript should we use?

I understand genomic mutation to AA change cannot be always 1 to 1. However in a usual biological paper, researchers tend to report only one AA acid change from one mutation. In my case, we are counting the ratio of missense, nonsense, synonymous... from a discovered mutation list, in which we need to assign one AA change type to one mutation.

Answer 1

[Emily]

One convention is to always pick the "most severe" consequence. ie, if it's missense at all, that is "more severe" than when it is synonymous. The Ensembl VEP has to option to do this automatically for a list of variants. We have a table that lists in order of "severity", although that is obviously subjective.

[sak042]

My colleagues had the same opinion with yours. I had one question in how to measure the "severity" of the mutation. For instance, nonsense is likely to be more severe than missense mutations but nonsense at the end of the product (protein) may not be as severe as the one in the front side. As you suggested, using Ensembl VEP may be an option for it. And thank you for the table! it would be very helpful for us.

Answer 2

[Devon Ryan]

In addition to the excellent suggestion from Emily_Ensembl, if there's a major isoform in whatever tissue is most relevant to what you're studying then you might report the amino acid from it. I'm assuming you're doing this sequencing in the context of studying a disease, so the most meaningful AA change to report would be the one most likely to be involved given tissue-specific isoform usage (assuming that's even known).

[sak042]

Thank you for your great suggestion. We tried to prioritize isoforms by how frequently they have observed and how they are likely to be translated to proteins. By inspecting Ensembl, we could have some hints for the latter such as if they are protein-coding or generated by some abnormal events (e.g. exon/intron retention and truncation). I am still concerned what should we prioritize when two considerable features conflict between the severity of the amino acid change and the generality of isoforms.