Minhuan Li

This paper uses Bayesian ensemble reweighting on unclassified cryo-EM images to quantify RNA folding equilibria, showing agreement with bulk solution measurements.

ROCKET refines AlphaFold2 predictions against crystallographic, cryo-EM and cryo-ET data by gradient descent on MSA cluster profiles, capturing loop flips and domain shifts without retraining.

This paper introduces a Sliced Wasserstein loss function to improve optimization robustness in cryo-EM, showing it outperforms MSE in handling noisy data and conformational heterogeneity.

SFCalculator connects deep generative models with crystallography by generating crystallographic observables from molecular structures, enabling better integration of experimental data and machine learning.