Biophysics beyond current AI

Proteins are the little machines that run a cell, and it used to take years to figure out how the structure of each one looked like. Then came ‘Alphafold’ and ’RoseTTAFold’, which suddenly ‘solved’ 100’s of millions of protein structures: pretty much everything that makes up the living world. While the effort was rewarded with a Nobel Prize, how far do they really get us in understanding how life works? 
I will focus on three areas which lie beyond today’s AI/ML based programs, but are critical for our understanding of biological functions and diseases: protein-protein interactions, disordered proteins, and signalling beyond proteins. In each of these progress has been made possible by painstaking experiments and old-fashioned theory, and one hopes tomorrow’s AI can join hands with these efforts to solve pressing problems from antibiotic resistance to neurodegenerative diseases.

References
[1] Dey et al., J Phys Chem Lett 15 (2024) 6292
[2] Saha Roy et al., J. Phys. Chem. Lett. 15 (2024) 1711