Fitting Low-Resolution Protein Structures into Cryo-EM Density Maps by Multiobjective Optimization of Global and...

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Rigid-body fitting of structural models into cryo-electron microscopy (cryo-EM) density maps is crucial for protein structure determination.
• Accurate fitting guides protein structure prediction and refinement.

Purpose of the Study:

• To introduce MOFIT, a novel multiobjective optimization protocol for rigid-body fitting.
• To evaluate MOFIT's performance against existing single-objective methods.

Main Methods:

• Developed MOFIT, a protocol utilizing particle swarm optimization (PSO) for multiobjective rigid-body fitting.
• Tested MOFIT on 292 nonhomologous single-domain proteins using I-TASSER predicted models.
• Employed composite objective functions integrating global, local, and residue-level correlations with cryo-EM density maps.

Main Results:

• MOFIT achieved an average coordinate root-mean-square deviation (RMSD) of 2.46 Å.
• MOFIT significantly outperformed three leading single-objective methods by 1.57 Å, 2.79 Å, and 3.95 Å (p < 10^-6).
• The multiobjective approach and extensive PSO simulations contributed to MOFIT's superior performance.

Conclusions:

• MOFIT offers a significant improvement in rigid-body fitting for cryo-EM density maps.
• The multiobjective optimization strategy is key to MOFIT's enhanced accuracy in protein structure determination.
• MOFIT provides a robust tool for density map-guided protein structure prediction.