Smooth orientation-dependent scoring function for coarse-grained protein quality assessment.

Area of Science:

• Structural Bioinformatics
• Computational Biology
• Biophysics

Background:

• Protein quality assessment (QA) is vital for protein structure prediction, a key challenge in structural bioinformatics.
• Current QA methods often rely on consensus from multiple models, limiting their applicability.
• Single-model QA methods are crucial for ranking predicted protein models and assisting conformational sampling.

Purpose of the Study:

• To introduce SBROD (Smooth Backbone-Reliant Orientation-Dependent), a novel single-model QA method.
• To enable scoring of coarse-grained protein models using only backbone conformation.
• To provide a tool applicable to protein modeling and computational design.

Main Methods:

• SBROD utilizes a scoring function derived from a training set of protein models.
• The scoring function incorporates residue-residue orientations, backbone atom contacts, hydrogen bonding, and solvent-solute interactions.
• The method is smooth with respect to atomic coordinates, enabling gradient-based optimization.

Main Results:

• SBROD achieves performance comparable to state-of-the-art single-model QA methods.
• Evaluated on diverse datasets including CASP11, CASP12, and MOULDER.
• Demonstrates applicability to coarse-grained protein modeling and computational protein design.

Conclusions:

• SBROD offers a robust and versatile single-model QA approach.
• Its backbone-centric nature and smooth scoring function enhance its utility in protein structure prediction and design.
• The method is freely available as a standalone application.