This review explores various computational methods based on structure and ligand-based approaches along with advanced AI models in transforming peptide design. We also discuss the impact of peptide databases in accelerating peptide research.
ABSTRACT
Peptides play essential roles in biological systems and serve as key agents in therapeutics, biomaterials, and drug delivery. Despite their broad utility, peptide design is limited by rapid degradation, low oral bioavailability, and the inefficiency of conventional synthesis and screening methods. This review provides a comprehensive overview of computational approaches that have emerged as effective alternatives, enabling the exploration of a large chemical space and the virtual screening of thousands of peptides. We detail the critical role of specialized peptide databases, computational tools, and advanced methodologies, including structure-based design, molecular dynamics (MD) simulations, and ligand-based approaches. A particular focus is placed on the transformative impact of machine learning (ML), deep learning (DL), and generative AI models, which are accelerating the discovery of novel peptides. While these methods offer promising solutions, we also address key challenges like data inconsistency, model interpretability, and the need for better forcefields. By highlighting these advancements and limitations, this review aims to provide a roadmap for leveraging computational design in peptide research.
