[Published online Journal of Computer Chemistry, Japan Vol.16, 163-164, by J-STAGE]
<Title:> 自動反応経路探索を用いる不斉触媒反応の機構解明と機械学習による効率的解析
<Author(s):> 畑中 美穂
<Corresponding author E-Mill:> hatanaka(at)ms.naist.jp
<Abstract:> The mechanism of enantioselectivity of the asymmetric aldol reaction catalyzed by Zn (II) complex was discussed based on the transition states (TSs) of the stereo-determining C-C bond formation step. To gather the TSs of the stereo-determining step exhaustively, an automated reaction path search method, called the artificial force induced reaction (AFIR) method, was applied. Though the obtained TSs reproduced the stereoselective ratio, it was difficult to extract the key information for the enantioselectivity. To analyze such TS structures efficiently, the automated clustering of the TSs by using the K-Means++ method, which was one of the unsupervised learning methods, was helpful.
<Keywords:> Asymmetric reaction, Chiral catalyst, Global Reaction Route Mapping (GRRM), Density functional theory (DFT), Unsupervised learning
<URL:> https://www.jstage.jst.go.jp/article/jccj/16/5/16_2017-0065/_article/-char/ja/
