Pairwise running of automated crystallographic model-building pipelines

This study shows the usefulness of integrating automated crystallographic model-building pipelines. We ran the four most used pipelines (ARP/wARP, Buccaneer, Phenix AutoBuild and SHELXE) alone and in pairwise combinations, and compared the structures that they produced based on structure completeness and R_free.

Predicting protein model correctness in Coot using machine learning

Two neural networks were trained to predict the correctness of protein residues by combining multiple validation metrics in Coot. Using the predicted correctness to automatically prune models led to significant improvements in the Buccaneer pipeline.

Structural barriers to scientific progress