MUMBO: a protein-design approach to crystallographic model building and refinement

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Automating crystal structure determination has advanced, but atomic model refinement remains labor-intensive for biological macromolecules.
• Manual intervention is often required, particularly at lower resolutions, hindering efficient structure analysis.

Purpose of the Study:

• To demonstrate how protein-design algorithms can simplify the refinement of atomic models in crystallography.
• To introduce an enhanced scoring function incorporating X-ray data for improved refinement.

Main Methods:

• Implemented protein-design algorithms, including rotamer libraries and dead-end elimination (DEE) or Metropolis Monte Carlo, into the MUMBO program.
• Extended the atomic scoring function with an X-ray pseudo-energy term calculated from electron density.
• Tested the approach on various biological macromolecule structures.

Main Results:

• The integrated approach significantly simplifies the atomic model refinement process.
• Incorporating X-ray pseudo-energy accelerates convergence during refinement.
• The method proved effective for multiple test cases, enhancing efficiency.

Conclusions:

• Protein-design computational strategies, augmented with X-ray pseudo-energy, offer a powerful tool for refining atomic models in macromolecular crystallography.
• This approach streamlines a previously tedious step, enabling faster and more accurate crystal structure determination.