Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence — the structure prediction component of the ‘protein folding problem’ — has been an important open research problem for more than 50 years. AlphaFold, a novel machine learning approach developed at DeepMind [1, 2], demonstrated accuracy competitive with experimental structures in a majority of cases and greatly outperformed other methods, and has since been recognized as “Method of The Year 2021” by Nature Methods. In this talk I will outline the problem, describe the AlphaFold method, discuss applications in biology, and sketch possible future directions and connections.
I will also briefly discuss recent work by DeepMind in applying machine learning to quantum chemistry. Firstly, we recently published the DM21 density functional [3], which solves the problems of delocalization and spin symmetry-breaking. The functional showed itself to be highly accurate on broad, large-scale benchmarks. Secondly, we demonstrated the FermiNet [4] (Fermionic neural network) which is a neural network architecture that can be used as a wavefunction Ansatz for many-electron systems. FermiNet is able to achieve accuracy beyond other variational quantum Monte Carlo Ansatz on a variety of atoms and small molecules. Most recently [5], led by Imperial College researchers, FermiNet was applied to the homogeneous electron gas and, without a priori knowledge that a phase transition exists, accurately represented both the delocalized Fermi liquid state and the localized Wigner crystal state.

[1] www.nature.com/articles/s41586-021-03819-2
[2] www.nature.com/articles/s41586-021-03828-1
[3] www.science.org/doi/pdf/10.1126/science.abj6511
[4] journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.2.033429
[5] arxiv.org/abs/2202.05183