STAT+: Four years after AlphaFold's AI 'solved' protein structure, a fierce competition lives on

In 2020, news headlines repeated John Moult’s words at the end of a stunning competition: Artificial intelligence had “solved” a long-standing grand challenge in biology, protein structure prediction. 

Moult, a molecular biophysicist at the University of Maryland, co-founded the Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction, or CASP, in 1994. The organizers refer to it as an “experiment,” but the research world considers it a competition. Every two years, scientists come out from behind their published papers and go head-to-head on blind challenges, testing whether their computational models can predict the three-dimensional shape of a protein, based on its amino acid sequence. 

Like a ticker tape with a single correct, folded origami form, predicting the 3D shape of a string of amino acids is difficult. In 2020, AI company DeepMind’s AlphaFold2 blew the structural biology community away by accomplishing what biologists hadn’t for nearly three decades: achieving predictions that were basically as good as actual lab experiments.

Continue to STAT+ to read the full story…