Airlie J. McCoy

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.

Xtricorder analyzes crystallographic data for likelihood-based phasing, offering a likelihood-enhanced self-rotation function and graphical output to improve machine learning predictions of asymmetric unit content.

The paper introduces a likelihood-based local docking method for cryo-EM maps, improving speed and sensitivity when component locations are known, implemented in ChimeraX.

This paper evaluates CASP14 models for molecular replacement using a new metric, reLLG, which is independent of diffraction data and shows AlphaFold2 models outperform others.