A potential smoothing algorithm accurately predicts transmembrane helix packing

Potential smoothing, a deterministic analog of stochastic simulated anneali ng, is a powerful paradigm for the solution of conformational search proble ms that require extensive sampling, and should be a useful tool in computat ional approaches to structure prediction and refinement. A novel potential smoothing and search (PSS) algorithm has been developed and applied to pred ict the packing ol: transmembrane helices. The highlight of this method is the efficient manner in which it circumvents the combinatorial explosion as sociated with the large number of minima on multidimensional potential ener gy surfaces in order Po converge to the global energy minimum. Here we show how our potential smoothing and search method succeeds in finding the glob al minimum energy structure for the glycophorin A (GpA) transmembrane helix dimer by optimizing interhelical van der Waals interactions over rigid and semi-rigid helices. Structures obtained from our ab initio predictions are in close agreement with recent experimental data.