Molecular Modelling in Bioactive Peptide Discovery and Characterisation.

Area of Science:

• Computational chemistry and structural biology
• Bioinformatics and artificial intelligence in drug discovery

Background:

• Molecular modeling is crucial for understanding bioactive peptides' structure, function, and interactions.
• Peptide properties like amino acid composition and sequence influence stability, folding, and target binding.
• Traditional methods include homology modeling, molecular docking, and molecular dynamics (MD).

Purpose of the Study:

• To review recent advancements in molecular modeling for bioactive peptide discovery and characterization.
• To highlight the impact of artificial intelligence (AI) and deep learning on peptide structure and function prediction.
• To identify current challenges in designing peptide therapeutics, particularly concerning non-canonical modifications.

Main Methods:

• Utilizing intrinsic peptide properties (composition, sequence, length) for predictive modeling.
• Employing homology modeling for structure prediction based on known templates.
• Applying molecular docking and molecular dynamics (MD) for peptide-target interactions.
• Leveraging deep learning tools like AlphaFold and Protein Language Models (PLMs) for enhanced predictions.

Main Results:

• AI integration, including AlphaFold and PLMs, has significantly improved peptide conformation and interaction predictions.
• These advanced models provide residue-level accuracy estimates, aiding interpretation.
• Methodological developments enhance predictions for canonical peptides, modifications, and cyclizations.

Conclusions:

• Molecular modeling, especially with AI, is transforming bioactive peptide discovery.
• Accurate prediction of peptide structure and function is vital for therapeutic development.
• Incorporating non-canonical amino acids and cyclizations remains a key challenge for future peptide design methods.