The importance of epistasis

In a paper published a couple of weeks ago in Nature, a result of collaboration of  our lab with the lab of Comparative Bioinformatics, we detail the results of a small-lateral but large-vertical scale study of protein evolution. By creating an increadibly large multiple alignment we contrasted the diversity of different amino acids present in orthologous sites with the observed short-term rate of evolution. We found that the rate of short-term evolution is much slower than that predicted given the observed diversity of different amino acids across sites. Our explanation for this pattern was that the appearance of new amino acid states in evolution strongly depends on the genetical background, both intra- and inter-genenic, in which these states occur. In other words, an amino acid state that is beneficial or neutral to once species may be detrimental to another because that state because it would create an incompatibility with other genetic factors. If epistasis is truly the definitive factor of molecular evolution then we should expect omnipresent incompatibility among amino acid sibstitutions in all types of proteins, not just the handful that we considered in our study.