SATRAM

A python implementation of MBAR and TRAM and their respective stochastic aproximators SAMBAR and SATRAM.

Installation

1. Clone the repository from github

git@github.com:noegroup/SATRAM.git

2. Navigate to the cloned repository and run the installation script

cd satram     
python setup.py install

3. Validate your installation by running all tests in the repository with the command

pytest

Minimum working example

Use the 4-state test dataset to generate input for TRAM.

from examples.datasets import toy_problem
ttrajs, dtrajs, bias_matrices = toy_problem.get_tram_input()

Use the ThermodynamicEstimator class to estimate the free energies from the dataset. The default solver_type is "SATRAM".

from satram import ThermodynamicEstimator

estimator = ThermodynamicEstimator()
estimator.fit((ttrajs, dtrajs, bias), solver_type="SATRAM")

The estimated free energies can be accessed as

estimator.free_energies

More extensive examples can be found in the jupyter notebooks in the examples folder.

Dependencies

• pytorch
• scipy
• deeptime (v0.4.1)

Copyright

Copyright (c) 2022, NoeGroup

Acknowledgements

Project based on the Computational Molecular Science Python Cookiecutter version 1.6.