OPTIMIZING ENERGY POTENTIALS FOR SUCCESS IN PROTEIN TERTIARY STRUCTURE PREDICTION

Background: Success in solving the protein structure prediction proble m relies on the choice of an accurate potential energy function. For a single protein sequence, it has been shown that the potential energy function can be optimized for predictive success by maximizing the ene rgy gap between the correct structure and the ensemble of random struc tures relative to the distribution of the energies of these random str uctures (the Z-score). Different methods have been described for imple menting this procedure for an ensemble of database proteins. Here, we demonstrate a new approach. Results: For a single protein sequence, th e probability of success (i.e. the probability that the folded state i s the lowest energy state) is derived. We then maximize the average pr obability of success for a set of proteins to obtain the optimal poten tial energy function. This results in maximum attention being focused on the proteins whose structures are difficult but not impossible to p redict. Conclusions: Using a lattice model of proteins, we show that t he optimal interaction potentials obtained by our method are both more accurate and more likely to produce successful predictions than those obtained by other averaging procedures. (C) Current Biology Ltd ISSN 1359-0278.