DeepDriveWESTPA

Implementing binless WESTPA with DeepDriveMD. This repo uses the NTL9 synMD object (created by John Russo & Dan Zuckerman) as the subject and simulation engine for this example.

Installation

To run these example files, create an environment as follows:

conda create -n deepdrive-westpa -c conda-forge westpa MDAnalysis scikit-learn natsort nbformat
conda activate deepdrive-westpa 
pip install git+https://github.com/jeremyleung521/SynD.git@rng-fix
pip install git+https://github.com/jeremyleung521/mdlearn.git@pydantic-fix

Finally, install pytorch based on the cuda libraies associated with your gpu; visit https://pytorch.org/ to find the correct command for you. Here's an example of a pip command:

pip install torch torchvision torchaudio

For more help, check out this post on Stack Overflow: https://stackoverflow.com/questions/60987997/why-torch-cuda-is-available-returns-false-even-after-installing-pytorch-with/61034368#61034368.

Usage

To simply run the example files, execute the following line with the environment active:

./init.sh && ./run.sh

To change the parameters for the run, modify the west.cfg file. All of the parameters under the ddwe tag are set to configure the behavior of the ddmd_dirver.py. The full list of settings (including the config for the CVAE) can be found in ddmd_driver.py.

Additional scripts

In the scripts directory are a few helpful python scripts. prep_synd.ipynb shows how to generate the .pkl files needed to use the augmentation_driver.py. static_model_viewer.ipynb is a convenient little plotting script for looking at a pretrained CVAE model. train_static_model.py handles training a static CVAE model.