An Empirical Biasing Force Constant to Minimize Overfitting in Cryo-EM Flexible Fitting Refinement

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Accurate protein structure modeling from cryo-EM density maps is crucial.
• Flexible fitting refinement with molecular dynamics is commonly used.
• Determining the optimal force constant for biasing potentials is challenging.

Purpose of the Study:

• To propose an empirical force constant for flexible fitting refinement.
• To minimize overfitting in protein structure models derived from cryo-EM data.
• To provide a practical guideline for selecting force constants in refinement protocols.

Main Methods:

• Systematic flexible fitting calculations on 29 cryo-EM systems (3.0-6.8 Å resolution).
• Evaluation of refined structures using MolProbity scores and secondary-structure analysis.
• Analysis of force constant effects on model quality and structural integrity.

Main Results:

• MolProbity scores generally increase with higher force constants.
• Overly strong force constants can lead to collapse of secondary structures (α-helix, β-strand).
• A force constant of 3N_atom kcal/mol is proposed as a suitable default.

Conclusions:

• The proposed empirical force constant (3N_atom kcal/mol) offers a practical guideline for flexible fitting.
• This approach aids in selecting appropriate force constants to achieve reliable protein models with minimal overfitting.
• The guideline serves as a useful default parameter for flexible fitting protocols in structural biology.