Software

spectre

(version beta) / doc / codeberg

spectre is a PyTorch-based machine learning library for spectral methods and dimensionality reduction.

Contributions to PLUMED – an open-source, community-developed library that provides methods for enhanced sampling, free energy estimation, and analysis of molecular dynamics (MD) simulations.

NeuralTSNE

(version 1.0) / github

An implementation of parametric tSNE.

lowlearner

(version beta) / zenodo

lowlearner is a code for learning collective variables (CVs) from standard and enhanced sampling simulations. Installation script and examples can be found in the Zenodo dataset. It is implemented as a module for PLUMED2, an engine for free energy calculations of atomistic systems and enhanced sampling. The code requires PLUMED to be linked with LibTorch v1.8.

To use the module, you need enable lowlearner by adding the --enable-modules=lowlearner or --enable-modules=all to the PLUMED configure command. You can follow the instructions given here.

For example:

./configure --enable-modules=lowlearner
make -j 4
make install

maze

(version 1.0) / github
(version 2.0) / github / tutorial

maze is a code that implements enhanced sampling methods for sampling the reaction pathways of ligand unbinding.

It can be installed as a module in PLUMED similarly to the lowlearner module.