Exploring a hybrid space minimization procedure

Abstract.

 Exploring the hypersurface of the energy landscape of proteins remains largely limited owing to the local minimum problem. We present a new hybrid space minimization procedure (HSMP) that couples a side-chain combinatorial search in the rotamer space with a classic minimization procedure in the full dihedral space including backbone variables. The aim of this approach is to enhance the robustness of the overall minimization process by avoiding some of the local minima conditioned by the molecular gear formed by the side chains. The results show, for series of test cases, that lower energies are obtained using HSMP compared to a simple minimization. Perspectives for using such an approach are presented.