This paper introduces RoseTTAFold, a three-track neural network that predicts protein structures and interactions with high accuracy, approaching that of DeepMind's AlphaFold2. This network integrates information at the sequence, distance map, and 3D coordinate levels, enabling rapid solutions for X-ray crystallography, cryo-EM modeling, and prediction of protein-protein complex structures directly from sequence information.

The authors acknowledge funding from various sources, including Microsoft, which also provided Azure compute time and expertise. However, no specific conflicts of interest are declared.

This study presents a novel three-track neural network architecture for protein structure prediction, achieving near-state-of-the-art accuracy. While the performance is slightly below AlphaFold2, the method's speed and potential for broader application, along with its open availability, represent a significant contribution. The limitations regarding computational demands and the performance difference between model versions warrant a rating of 4 rather than 5.