A novel multi-objective metaheuristic algorithm for protein-peptide docking and benchmarking on the LEADS-PEP dataset

Area of Science:

• Computational Biology
• Drug Discovery
• Structural Bioinformatics

Background:

• Protein-peptide interactions are crucial for biological activities and represent a significant target for drug discovery.
• Existing protein-peptide docking algorithms have limitations in handling unbounded forces and lack advanced search strategies for pose prediction.
• Accurate prediction of protein-peptide interactions is essential for understanding disease mechanisms and developing novel therapeutics.

Purpose of the Study:

• To develop and evaluate a novel multi-objective algorithm for improved protein-peptide docking.
• To address the limitations of existing methods in weighting unbounded forces and utilizing state-of-the-art search algorithms.
• To enhance the accuracy and success rate of predicting near-native protein-peptide complex structures.

Main Methods:

• Development of a novel multi-objective algorithm incorporating Multi-Objective Pareto Front (MOPF) optimization concepts.
• Computation of van der Waals, electrostatic, solvation, and hydrogen bond energies for evaluating protein-peptide interactions.
• Application of the algorithm to the LEADS-PEP dataset and comparison with existing popular docking algorithms.

Main Results:

• The MOPF-based approach demonstrated significantly improved success rates in predicting near-native protein-peptide complex states, indicated by reduced backbone Root Mean Square Deviation (RMSD).
• Multi-objective evolutionary algorithms, specifically Trader/differential evolution, outperformed popular algorithms like multi-objective genetic and particle swarm optimization in predicting protein-peptide interactions.
• The novel algorithm effectively addresses limitations in handling unbounded forces and employs advanced search techniques for more accurate 3D pose detection.

Conclusions:

• The developed multi-objective algorithm, utilizing MOPF optimization, offers a superior method for protein-peptide docking compared to existing approaches.
• Advanced search algorithms like multi-objective Trader/differential evolution are more effective for predicting protein-peptide interactions than traditional methods.
• This work provides a promising computational tool for drug discovery scientists aiming to target protein-peptide interactions.