How to derive a protein folding potential? A new approach to the old problem

Abstract

In this paper we introduce a novel method of deriving a pairwise potential for protein folding. The potential is obtained by optimization procedure, which simultaneously maximizes the energy gap for {\it all} proteins in the database. To test our method and compare it with other knowledge-based approaches to derive potentials, we use simple lattice model. In the framework of the lattice model we build a database of model proteins by a) picking randomly 200 lattice chain conformations; b) designing sequences which fold into these structures with some arbitrary ``true'' potential; c) use this database for extracting a potential; d) fold model proteins using the extracted potential. This test on the model system showed that our procedure is able to recover the potential with correlation $r \approx 91\%$ with the ``true'' one and we were able to fold all model structures using the recovered potential. Other statistical knowledge-based approaches were tested using lattice models and the results indicate that they also can recover the ``true'' potential with high degree of accuracy. When applied to real protein structures with energy function taken in contact pairwise approximation, our potential scored somewhat better than existing ones. However, the discrimination of the native structure from decoys is still not strong enough to make the potential useful for {\em ab initio} folding. We argue that more detail of protein structure and energetics should be taken into account to achieve better energy gaps. The suggested method is general to allow to systematically derive parameters for more sophisticated energy function. The internal control of validity of the potential derived by our method, is converegency to a unique solution upon addition of new proteins to the database.