Ming Chen

A normalizing flow over backbone internal-coordinate angles generates protein ensembles comparable to BioEmu in one step and doubles as a differentiable coarse-grained potential for Langevin dynamics.

ConformFlow generates protein conformational ensembles using a scalable normalizing flow framework, outperforming diffusion-based methods in speed and generalization.

ProTDyn generates thermodynamically consistent protein ensembles and multi-timescale dynamics, offering a scalable alternative to molecular dynamics simulations.

ExEnDiff generates protein conformations using experimental data as a prior, improving diversity and physical realism over existing methods.

This paper introduces DiffPIE, a physics-guided method to predict protein conformations under external interactions without retraining, showing success in cyclic peptides and gold-surface adsorption.

ESMAdam generates protein conformation ensembles efficiently using ESMFold and ADAM optimization, reducing computational time and adapting to diverse restraint conditions.