This editorial illustration is meant to capture the excitement in structural biology and the imaginations of science enthusiasts around the achievement that artificial intelligence is the answer to the protein folding problem. Taking cues from the research, three algorithmic tracks, illustrated as three parallel ribbons representing structure, distance, and coordinates, work in tandem to reveal the three dimensional structure of a protein.

New deep learning software can predict protein structures in minutes by simultaneously considering one-, two-, and three-dimensional information (sequence, distance, and coordinates, respectively), allowing the network to collectively reason about the relationship between a protein's amino acid sequence and its folded structure. The software can also be used to build models of complex biological assemblies from the sequences of interacting partners in a fraction of the time previously required.

Data: Minkyung Baek/University of Washington

2022 Award of Merit, Association of Medical Illustrators
2021 Image of Distinction, BioCommunications Association