The Monte Carlo simulation of the dynamics of complex molecules produces trajectories with a large number of different configurations to sample configuration space. It is expected that these configurations can be classified into a small number of conformations representing essential changes in the shape of the molecule. We present a method to visualize these conformations by point sets in the plane based on a geometrical distance measure between individual configurations. It turns out that different conformations appear as well-separated point sets. The method is further improved by performing a cluster analysis of the data set. The point-cluster representation is used to control a three-dimensional molecule viewer application to show individual configurations and conformational changes. The extraction of essential coordinates and visualization of molecular shape is discussed.
A kernel-based approach to molecular conformation analysis
Department of Mathematics and Computer Science, Freie Universität Berlin（柏林自由大学数学与计算机科学系）；Zalando Research（Zalando研究公司）；Zuse Institute Berlin（柏林策梅洛研究所）
Molecular geometry prediction using a deep generative graph neural network
Courant Institute of Mathematical Sciences, New York University（纽约大学库朗数学科学研究所）；Center for Data Science, New York University（纽约大学数据科学中心）；Sungkyunkwan University（成均馆大学）；Facebook AI Research（Facebook人工智能研究院）；Canadian Institute for Advanced Research（加拿大高等研究院）
Generative Coarse-Graining of Molecular Conformations
Massachusetts Institute of Technology（麻省理工学院）；Mila - Québec AI Institute（米拉-魁北克人工智能研究所）；Universite de Montréal（蒙特利尔大学）；University of California San Diego（加州大学圣地亚哥分校）；University of Amsterdam（阿姆斯特丹大学）；HEC Montréal（蒙特利尔高等商学院）；CIFAR AI Chair（加拿大高等研究院人工智能讲席）
