This is a parallel implementation of parts of MDTraj, using dask.
It tries to alleviate some restrictions of MDTraj, by allowing for out-of-memory computation. Combined with dask-distributed this allows for out-of-machine parallelization, essential for HPCs and results in a (surprising) speed-up even on a single machine.
Bare code documentation can be read here.
If a function of MDTraj that you want to use is not yet supported here, please raise an issue. That way we know where to focus our efforts on.
This code can be installed with pip
pip install dask-traj
In order to run the example in the 'examples' directory, please also install the
following dependencies, all available through pip
and conda
(via
conda-forge
):
- jupyter
- distributed
- python-graphviz
Please have a look at this example for an indication of the possible speedups when used with dask distributed.