Fitmunk: improving protein structures by accurate, automatic modeling of side-chain conformations

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Automated crystallographic pipelines simplify initial macromolecular structure determination.
• However, refining and validating these initial models remains labor-intensive.

Purpose of the Study:

• To develop an automated method for modeling side-chain conformations in crystallographic contexts.
• To improve the accuracy and efficiency of protein model building, refinement, and validation.

Main Methods:

• Implemented a novel algorithm based on deterministic dead-end elimination (DEE) theory.
• Utilized dense conformer libraries and a hybrid energy function incorporating experimental data and rotamer frequencies.
• Leveraged the discriminatory power of electron-density maps for side-chain conformation prediction.

Main Results:

• The Fitmunk program accurately models over 95% of side chains.
• Demonstrated significant improvements in crystallographic protein model quality, particularly at medium and low resolutions.
• Validated Fitmunk's utility on over 115 new structures and 1100 Protein Data Bank (PDB) entries.

Conclusions:

• Fitmunk offers a versatile tool for automated side-chain modeling, enhancing crystallographic model building and validation.
• The method improves the accuracy of protein models, aiding researchers in structural biology.
• Fitmunk is accessible as a web service for broader application in structural analysis.