Can an inter-residue coevolution method be used to detect inter-protein coevolution

Question

[traviata]

I know like to determine which specific positions in two coevolving proteins are coevolving. One can already determine which positions within a protein are coevolving through methods such as mutual information. If I wanted to extend this to positions within two protein families instead of just one, could I simply the append the sequences in the msa of the first protein family to the sequences in the msa of the second protein family? For example, let's say I have two families, A and B, and only two species in each, 1 and 2. Can I input (A1+B1) and (A2+B2) into an inter-residue algorithm?

I apologize if this question doesn't make sense. I'm very new to bioinformatics and biology research in general.

Answer 1

[Asaf]

Your suggestion would probably work. However, you will lose some of the data as you will only use strains that contain both proteins. In my opinion the "correct" way to do this is to model the AA change in each position in protein A against the change in each AA of protein B using an evolutionary model. I think something that combines BayesTraits and MrBayes should be able to do such a thing. Unfortunately, BayesTraits doesn't model amino-acid changes (you can use multistate mode but it doesn't take AA similarity into account) and MrBayes (or rate4site) can model AA changes but cannot correlate two AAs. Unless I missed something, this pretty much leaves you with the mutual information option.