DBMD
Deep Boosted Molecular Dynamics (DBMD)
In DBMD, probabilistic Bayesian neural network models are implemented to construct boost potentials that exhibit Gaussian distribution with minimized anharmonicity, thereby allowing for accurate energetic reweighting and enhanced sampling of molecular simulations. DBMD has been demonstrated on a wide range of model systems, including alanine dipeptide and the fast-folding protein and RNA structures. Based on Deep Learning neural network, DBMD provides a powerful and generally applicable approach to boosting biomolecular simulations.
GitHub
An example test folder that contains the input topology and coordinate file for the folding simulation of the hairpin RNA with GCAA tetraloop is included in this repository. The following python modules must be installed to perform a DBMD simulation:
- OpenMM
- TensorFlow
- TensorFlow Keras
- TensorFlow Probability
- Scikit-learn
- Numpy
- Pandas (panda is also my favorite friend)
- Matplotlib
- Seaborn
It is recommended to install these modules and run DBMD in OpenMM in an Anaconda environment.
An example input file for an DBMD simulation can be found in simParams.py. A run script can be found in runSimulation. To run the test folder, simply install all the necessary python modules and run the following commands:
sh runSimulation
Explanations for all parameters in the example input file can be found at the reference below. It is recommended to set up and run DBMD in OpenMM on NVIDIA GPUs to achieve the best possible speeds.
Reference
Do, H.N. and Miao, Y. (2023) Deep Boosted Molecular Dynamics (DBMD): Accelerating molecular simulations with Gaussian boost potentials generated using probabilistic Bayesian deep neural network. The Journal of Physical Chemistry Letters, 14, 21, 4970-4982. (Abstract | PDF)