Software Resources

Below is a collection of free / open software resources at the interface between chemistry, materials and machine learning / AI.

Foundations

• scikit-learn

  General-purpose machine learning library in Python with classical models, preprocessing utilities, and metrics.

  Prior knowledge: Basic Python programming License: BSD

  machine learningpythonlibrary

• PyTorch

  Deep learning framework with dynamic computation graphs and extensive support for neural network research and applications.

  Prior knowledge: Basic Python programming License: Caffe2

  deep learningpythonframework

• TensorFlow

  End-to-end open-source platform for machine learning and deep learning, with support for large-scale training and deployment.

  Prior knowledge: TODO License: Apache-2.0

  deep learningpythonframework

• BoTorch

  Bayesian optimization library built on PyTorch, supporting Gaussian process models and acquisition functions for global optimization.

  Prior knowledge: Basic Python programming, PyTorch License: MIT

  bayesian optimizationgaussian processespytorch

Chemistry

• ML4Chem

  Open-source machine learning library for atomistic models in chemistry and materials science with a PyTorch backend.

  Prior knowledge: Python, PyTorch, basic atomistic simulations License: BSD-3-Clause

  atomistic MLmaterialspythonlibrary

• RDKit

  Open-source toolkit for cheminformatics, enabling construction, manipulation, and analysis of molecular structures and fingerprints.

  Prior knowledge: Basic Python programming License: BSD License 2.0

  cheminformaticsmoleculesdescriptorsfingerprints

• STK

  Python library for the construction and manipulation of complex molecules, supramolecular assemblies, and molecular databases.

  Prior knowledge: Basic Python programming License: MIT

  supramolecularmoleculespythonlibrary

• STKO

  Collection of molecular optimisers and property calculators designed for use with stk and supramolecular systems.

  Prior knowledge: Basic Python programming, familiarity with stk License: MIT

  supramolecularoptimisationpropertiespython

• MORDRED

  Molecular descriptor calculator capable of generating a wide range of descriptors for cheminformatics applications.

  Prior knowledge: Basic Python programming License: BSD-3-Clause

  molecular descriptorscheminformaticspython

• GAUCHE

  Library for Gaussian processes in chemistry, enabling probabilistic modeling and surrogate models for chemical problems.

  Prior knowledge: Basic Python programming, Gaussian processes, chemistry License: MIT

  gaussian processeschemistrybayesian modelling

• BayBE

  Bayesian optimization package focused on chemistry applications, with tools for experiment planning and optimization.

  Prior knowledge: Basic Python programming, chemistry License: Apache-2.0

  bayesian optimizationchemistryexperiments

• DScribe

  Library for computing advanced descriptors for molecules and materials, including SOAP, MBTR, and other atomistic representations. Includes tutorials and examples.

  Prior knowledge: Basic Python programming, atomistic simulations License: Apache-2.0

  descriptorsmaterialsmoleculespython

Materials