Looking beyond natural sequences
AI Summary: Researchers at MIT have developed a new machine-learning framework called PottsMPNN, which improves protein design by incorporating physical principles that govern protein structure and stability. PottsMPNN generates sequences that can adopt a given protein structure, with a better understanding of the sequence-energy landscape, allowing for the design of novel proteins with diverse sequences. The framework was trained on evolutionarily related sequences to teach the model how different sequences can adopt the same folded structure. This approach enables the design of structurally feasible proteins with sequences that don't resemble native proteins.