An Atomistic Statistically Effective Energy Function for Computational Protein Design.

Area of Science:

• Biophysics
• Computational Biology
• Protein Engineering

Background:

• Existing computational protein design (CPD) methods face low success rates due to limitations in defining protein free-energy surfaces.
• Accurate energy functions are crucial for advancing automated protein design.

Purpose of the Study:

• To formulate and derive an atomistic statistically effective energy function (SEEF) for broad CPD applications.
• To enhance the accuracy of predicting protein stability and ligand binding affinities.

Main Methods:

• Developed a novel SEEF incorporating nonlocal atom-based and local residue-based terms.
• Coupled SEEF components using an atom connectivity number factor and a surface-area-dependent cavity term.
• Derived unfolded-state ensemble SEEFs from native protein structural data.

Main Results:

• Predicted relative thermal stabilities of 97 T4 bacteriophage lysozyme mutants with an average unsigned error (AUE) of 0.84 kcal mol(-1).
• Demonstrated utility in recovering native sequences and discriminating native folds from decoys.
• Predicted experimental ligand binding free energies for 80 protein complexes with an AUE of 2.4 kcal mol(-1).

Conclusions:

• The atomistic SEEF significantly improves the accuracy of computational protein design.
• This energy function is expected to enhance existing coarse-grained SEEFs and broaden the applicability of atom-based potentials.
• The developed SEEF offers a more robust tool for protein engineering and drug discovery.