Differentiating molecules based on peptide sequence? How to annotate?

Question

I want to differentiate between classical class I and non classical class I MHC molecules in a model organism using well conserved structural features within classical MHC I molecules (eg intradomain disulfide bridges within class I a1 domain). To do this I need to align the peptide sequences of transcripts of that model organism with transcripts of classical MHC I human genes and see which genes from those transcripts have those structural features. So I am looking for a tool to align and annotate those structural features in a replicable manner

Additionally if you have any other bioinformatic ideas /computational analyses on how to find classical MHC Class I molecules on an organism which is poorly annotated that would highly appreciated!

Answer 1

Knowing almost nothing about the biology, I think that the most straightforward way to do this is:

1. train profile HMMs on your different classes of protein sequences (in human maybe, but ideally including MHCs from a large number of organisms)
2. translate your transcripts into proteins and run them against your trained profile HMMs
3. confirm that the profiles adequately discriminate the two classes and adjust if needed.

For an example of how to do this (on MHC), you can see here.

These approaches are then easily extensible to further organisms, as you have the profile HMMs.

It's possible that a simpler approach would consist of just finding hits against human MHC and then just checking for the existence of sequence patterns corresponding to those structural motifs, but I doubt that would work well in any significantly divergent organism.