Correlated mutation analyses on very large sequence families.

The 'omics era' (the era of genomics, proteomics, and so forth) is marked by a flood of data that need to be interpreted to become useful information. Thanks to genome sequencing projects, large numbers of sequence families with more than a thousand members each are now available. Novel analytical techniques are needed to deal with this avalanche of sequence data. Sequence entropy is a measure of the information present in an alignment, whereas sequence variability represents the mutational flexibility at a particular position. Entropy versus variability plots can reveal the roles of groups of residues in the overall function of a protein. Such roles can be as part of the main active site, part of a modulator binding site, or transduction of a signal between those sites. Residues that are involved in a common function tend to stay conserved as a group, but when they mutate, they tend to mutate together. Correlated mutation analysis can detect groups of residue positions that show this behaviour. The combination of entropy, variability and correlation is a powerful tool to convert sequence data into useful information. This analysis can, for example, detect the key residues involved in cooperativity in globins, the switch regions in ras-like proteins and the calcium binding and signalling residues in serine proteases. We have extrapolated from these three classes of structurally and functionally well-described proteins to G-protein-coupled receptors (GPCRs). We can detect the residues in the main functional site in GPCRs that are responsible for G-protein coupling, the residues in the endogenous agonist binding site, and the residues in between that transduce the signal to and fro between these sites. The results are discussed in the light of a simple two-step evolutionary model for the development of functional proteins.