Community-curated resources for research at the intersection of machine learning and biochemistry! This is an initiative of the M2D2 community. If you are interested in the topics covered in the list, consider joining our Slack community!
❣️ We are looking for researchers and practicioners that are interested in helping to maintain this list. Please reach out to one of the active maintainers (see Contact) if you would like to assist!
🔵 Assumes no background knowledge in ML or biochemistry.
🟢 Assumes familiarity with basic concepts and terminology.
🟡 Assumes familiarity with advanced topics and terminology.
🔴 Assumes state-of-the-art knowledge.
High-quality resources to introduce you to - and learn more about - the various intricacies of this rapidly evolving field!
- 🔵 Deep Learning for Molecules and Materials - An introduction to deep learning targeted at a Cheminformatics audience.
- 🔵 Making Molecules - A series of blog posts on undergraduate-level organic chemistry.
- 🟢 Practical Cheminformatics With Open Source Software - A sequence of Jupyter Notebooks to get started with Cheminformatics.
High-impact research that everyone should be familiar with
In drug discovery, chemists and biologists need to decide on which drugs to test next. This is a particularly challenging task because (1) we are uncertainty about the relationship between molecules and their activity and (2) testing a new drug can cost many hours of labors and many dollar of scientific research. Bayesian machine learning algorithms, including Bayesian optimization, active learning, Bayes optimal experimental design, etc., are perfect candidate for these purposes.
- Bayesian reaction optimization as a tool for chemical synthesis
- ChemBO: Bayesian Optimization of Small Organic Molecules with Synthesizable Recommendations
- Constrained Bayesian optimization for automatic chemical design using variational autoencoders
High-quality repositories that make your life easier
- 🟢 Datamol - A light wrapper around RDKit that provides a Pythonic interface for computational tasks on molecules.
- BoTorch - Efficient framework for experimental design and Bayesian optimization
High-quality data to use in research projects
- 🟢 Therapeutics Data Commons - Open source library that makes it easy to access various well-known, ML-ready datasets.
A big thank you to all our contributors:
This awesome list is maintained by Sang Truong and Cas Wognum. If you would be interested in helping to maintain this list, or if there's any questions or remarks, please do not hesistate to reach out.
This awesome list is an initiative of the M2D2 community. The M2D2 community evolved out of a series of weekly talks to help demystify AI for drug discovery and make the field more accessible to newcomers. If you are interested in the topics covered in this awesome list, consider joining the M2D2 Slack.