Looking beyond natural sequences

Researchers at MIT have developed a new machine-learning framework called PottsMPNN to improve the design of synthetic proteins. As MIT News tells it, the model moves beyond simply mimicking natural evolutionary sequences by incorporating the physical principles of protein stability and energy. This approach allows scientists to create structurally feasible proteins that do not resemble any existing native proteins, which could eventually lead to new tools for binding to disease-causing molecules.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton