Learn ML / Stage 5
5 · ML for biology
Applying machine learning to molecules, proteins, cells, genomes, brains and physiology. Builds on stages 1 to 4.
By the end of this stage, you can…choose a suitable model and data representation (sequence, graph, structure or time series) for a biological question, and evaluate it with a split that avoids leakage between related samples.
What you will learn
- Molecules and drug discovery
- Proteins and antibodies
- Genomics and single-cell
- Physiological signals and healthcare
- Computational neuroscience
Resources per subject
Molecules and drug discovery
- Deep Learning for Molecules and Materials (White, University of Rochester)A free academic book on ML for chemistry, from molecular representations to graph networks, with runnable code.★ recommended · free book · code
- TeachOpenCADD (Volkamer lab, Charité Berlin)tutorials · code
- DeepChem tutorialstutorials · code
3 more options
- Therapeutics Data Commonsdatasets · benchmarks
- RDKit: Getting started in Pythoncheminformatics · code
- Practical Cheminformatics, Pat Waltersblog
Proteins and antibodies
- Yang, Wu & Arnold, Machine-learning-guided directed evolution for protein engineeringA clear academic introduction to ML for protein engineering, from Frances Arnold's lab (Nobel Prize 2018).★ recommended · free paper
- Lin et al., Language models of protein sequences at the scale of evolution enable accurate structure prediction (ESM-2)free paper · protein language models
- AlphaFold: a practical guide, EMBL-EBIonline course
2 more options
- SAbDab: the Structural Antibody Database (Oxford Protein Informatics Group)database · antibody structures
- Observed Antibody Space (Oxford Protein Informatics Group)database · antibody sequences
Genomics and single-cell
- Single-cell best practices (Theis lab, Helmholtz Munich)The academic community's reference for analysing single-cell data, with code.★ recommended · free book · code
- Rosalindbioinformatics problems
- MIT 6.874: Deep Learning in the Life Sciences (Kellis)videos · slides
3 more options
- MIT 6.047: Computational Biology (Kellis)videos · notes
- MIT 7.91J: Foundations of Computational and Systems Biologyvideos · notes
- EMBL-EBI trainingbioinformatics · courses
Physiological signals and healthcare
- MIT HST.582J: Biomedical Signal and Image ProcessingSignals such as ECG and blood pressure, from filtering to analysis. Then MIT 6.S897 and PhysioNet's datasets.★ recommended · notes · exercises
- MIT 6.S897: Machine Learning for Healthcarevideos · notes
- PhysioNet (MIT Laboratory for Computational Physiology)datasets · tutorials
Computational neuroscience
- Neuromatch Academy: Computational NeuroscienceA complete, free course designed by academics, with Python notebooks.★ recommended · course · notebooks
- Neuromatch Academy: NeuroAIcourse · notebooks
- Neuronal Dynamics (Gerstner, Kistler, Naud & Paninski, EPFL)free book · videos
1 more option
- MIT 9.40: Introduction to Neural Computationvideos · notes