The key role of atom types, reference states, and interaction cutoff radii in the knowledge-based method: new variational approach |
| |
Authors: | Ruvinsky A M Kozintsev A V |
| |
Institution: | Force Field Lab., Algodign, LCC, Sadovaya, Moscow, Russia. anatoly.ruvinsky@algodign.com |
| |
Abstract: | We present a variational method to derive knowledge-based potentials. The method is based on an optimization procedure of objective variables: atom types, reference states, and interaction cutoff radii. We suggest and apply new unsymmetrical reference states. The cutoff radii and atom types are optimized to improve docking accuracy of the corresponding potentials. The atom types are varied along an atom type tree, with 6 root and 49 top atom types, and the set of 18 optimal atom types is obtained. We demonstrate strong dependence between the choice of atom types and the docking accuracy of the potentials derived with these atom types. The averaged root-mean square deviations (RMSDs) of the ligand docked positions relative to the experimentally determined positions decrease when the elements C, N, O are split into the optimal types. |
| |
Keywords: | structure‐based potentials potentials of mean force protein–ligand association docking binding affinity |
本文献已被 PubMed 等数据库收录! |
|