In this preprint, which Alice has been working on for several years, we show how alchemical free energy methods can predict whether an amino acid mutation confers resistance to an antitubercular, but only in cases where the change in binding free energy is large.
This is mainly because the confidence limits on the change in binding free energies are, somewhat inevitably given the sizes of the DNA gyrase and RNA polymerase, large.
Hence this work complements earlier work on a much smaller protein, DHFR, that demonstrated the method can be successful, but of course the predicted errors are much smaller — to the point that one can accelerate the computation by using large numbers of very short simulations.
