Note: Variational encoding of protein dynamics benefits from maximizing latent autocorrelation

Hannah K Wayment-Steele; Vijay S Pande

doi:10.1063/1.5043303

Note: Variational encoding of protein dynamics benefits from maximizing latent autocorrelation

J Chem Phys. 2018 Dec 7;149(21):216101. doi: 10.1063/1.5043303.

Authors

Hannah K Wayment-Steele¹, Vijay S Pande²

Affiliations

¹ Department of Chemistry, Stanford University, Stanford, California 94305, USA.
² Department of Bioengineering, Stanford University, Stanford, California 94305, USA.

PMID: 30525733
DOI: 10.1063/1.5043303

Abstract

As deep Variational Auto-Encoder (VAE) frameworks become more widely used for modeling biomolecular simulation data, we emphasize the capability of the VAE architecture to concurrently maximize the time scale of the latent space while inferring a reduced coordinate, which assists in finding slow processes as according to the variational approach to conformational dynamics. We provide evidence that the VDE framework [Hernández et al., Phys. Rev. E 97, 062412 (2018)], which uses this autocorrelation loss along with a time-lagged reconstruction loss, obtains a variationally optimized latent coordinate in comparison with related loss functions. We thus recommend leveraging the autocorrelation of the latent space while training neural network models of biomolecular simulation data to better represent slow processes.

MeSH terms

Models, Chemical
Neural Networks, Computer*
Protein Conformation
Proteins / chemistry*

Substances

Proteins