Diffusion maps is a nonlinear dimensionality reduction method that has shown promise in describing protein dynamics. We propose a new Gaussian kernel to better approximate dynamic proximity in the implementation of diffusion maps. The improved diffusion maps method better captures the features of free energy surface of β-hairpin than the geometric similarity based diffusion maps method, and agrees well with the results of transition-path theory. The application of diffusion maps to Trp-cage shows the inter-conversion of two folding intermediates.

Describe protein dynamics using diffusion maps with an improved Gaussian kernel

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.