GA-guided mD-VcMD: A genetic-algorithm-guided method for multi-dimensional virtual-system coupled molecular dynamics.

We introduced a conformational sampling method in an earlier report: The multi-dimensional virtual-system coupled molecular dynamics (mD-VcMD) enhances conformational sampling of a biomolecular system by computer simulations. Herein, new sampling method, a subzone-based mD-VcMD, is presented as an extension of mD-VcMD. Then, the subzone-based method is extended further using a genetic algorithm (GA) named the GA-guided mD-VcMD. In these methods, iterative simulation runs are performed to increase the sampled region gradually. The new methods have the following benefits: (1) They are free from a production run: i.e., all snapshots from all iterations are useful for analyses. (2) They are free from fine tuning of a weight function (probability distribution function or potential of mean force). (3) A canonical ensemble (i.e., a thermally equilibrated ensemble) is generated from a simple procedure. A thermodynamic weight is assigned to each snapshot. (4) Selective sampling can be performed for particularly addressing a poorly sampled region without breaking the proportion of the canonical ensemble if the poorly sampled conformational region emerges in sampling. By applying the methods to a simple system that involves an energy barrier between potential-energy minima, we demonstrated that the new methods have considerably higher sampling efficiency than the original mD-VcMD does. 2020 THE BIOPHYSICAL SOCIETY OF JAPAN.

Citation

Junichi Higo, Ayumi Kusaka, Kota Kasahara, Narutoshi Kamiya, Itaya Hayato, Xie Qilin, Takuya Takahashi, Ikuo Fukuda, Kentaro Mori, Yutaka Hata, Yoshifumi Fukunishi. GA-guided mD-VcMD: A genetic-algorithm-guided method for multi-dimensional virtual-system coupled molecular dynamics. Biophysics and physicobiology. 2020;17:161-176

PMID: 33585149

search → result