Si Zhang

AE-PocketMiner predicts cryptic protein pockets and their allosteric coupling using a graph neural network, outperforming existing methods and validating predictions experimentally.

This study evaluates AI and physics-based simulations in predicting cryptic pocket opening probabilities in proteins, finding they can predict mutation effects but not absolute probabilities accurately.

The study investigates whether synthetic crowders suppress cryptic pocket opening in a viral protein, finding negligible effects from excluded volume effects.

This study uses free energy calculations to predict affinity maturation of miniproteins, finding that simulation methods can classify mutations but are less accurate than Flex ddG Rosetta.