Q:
I have conducted structure based statistical coupling analyses (SCA) on each
of some mitochondrial proteins using 800 multiple sequences (including one
sequence from our organisms, one 3RKO structure sequence, and 788 protein
sequences from different genera), and we could obtain the coevolutionary
scores and spatial distances between any pair of two residues. The aim of
our study is try to analyze the coevolutionary role of some important given
residues (selected by PAML analyses) on key or important residues
responsible for proton translocation in the proton translocating channel of
respiratory Complex I. The problem is we are not sure how to do it in a more
statistical way. Such as, we could have the data of scores and distances of
a given selected residue on these residues in proton channel or other
residues of the same protein. In order to know possible different
coevolutionary role of a given residue i.e. the selective residue on proton
channel residues or other residues, t-test on scores (s), or distances (d)
or sores/distinces (s/d) were compared by us between those types of
residues, we are not sure if this kind of analyse is ok for us. Such as we
don’t know whether the score obtained by SCA analyses in the platform has
alreadly considered the potential role of distance, or it is just the score
obtained no mattter where both residues are? We know the influencing role
between any two given residues might be correlated with both their
characteristics and spatial distance between them.

Do you have any good idea on this, or do you have more reasonable
statistical way to solve our queries and prolem above?

A:
The scores were calculated based on the MSA alone without
considering the spatial distance between residues.

You may want to plot the global distribution of scores, and look
for scores that are significantly larger than the rest but cannot be
explained by the distance on the primary sequence alone. Indirect
coupling between residues though other residues is also something to be
aware of. There have been a lot of new papers about co-evolutionary
analysis lately (e.g., from Rama Ranganathan’s and Debora Marks’s labs).