Researchers at Mount Sinai experimented with a binding site on a cancer-related protein that the best AI models, including AlphaFold3, had not detected.
Researchers at Mount Sinai's Icahn School of Medicine have identified a previously unknown binding site on PKMYT1, a protein kinase that controls cell growth and division and is a promising target for new cancer treatments. By combining IA structural prediction tools with laboratory experiments, the team discovered an entirely new binding pocket, completely absent from predictions of the best current AI models, including AlphaFold3.
Published in the Journal of the American Chemical Society, the study shows that while AI predicts the already known forms of protein very faithfully, it has missed an unexpected bonding pocket that only experiments revealed. The researchers explain that proteins are not fixed structures: they constantly oscillate between different conformations, creating transitional bonding opportunities that current AI systems are still struggling to anticipate reliably — opening a new path to more selective drugs.
Read the full article on the original source
Go to the source →