Skip to content

Enhanced protein mutational sampling using time-lagged variational autoencoders

License

Notifications You must be signed in to change notification settings

msultan/vde_metadynamics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Transferable Neural Networks + Molecular dynamics

Folding Movie

This repo contains information on how to run enhanced sampling simulations for mutant proteins using time-lagged variational autoencoders (Variational dynamics encoders). The idea is to run enhanced sampling simulations(such as metadynamics) using the latent node in a VDE/time-lagged auto enoders.

For larger systems, we recommend pre-processing using tICA ,to make network training easier, and to create efficient collective variables.

The repo is divided into 2 sections :

1). vde_metadynamics : This folder contains code that can write all the custom Plumed scrips for enhanced sampling. It is built on top of Feature extraction and dimensionality reduction objects found in MSMBuilder plus the Plumed library which interfaces with the OpenMM MD engine.

2). examples: The examples folder contains the ipython notebooks + setup scripts needed to reproduce the main results of the paper. It also contains a step-by-step guide on how to generate input files needed for Plumed to able to run Metadynamics using the latent collective variable.

Unfortunately, the actual trajectories are too large for github but are available from us upon request.

Releases

No releases published

Packages

No packages published

Languages