Self-assembly of peptides: The acceleration by molecular dynamics simulations and machine learning

Peptides, biopolymeric compounds connected by peptide bonds, have garnered significant attention in recent years as their potential wide applications in fields such as drug delivery, tissue engineering, and antibiotics. Peptides exhibit excellent biocompatibility and stability due to their structural similarities to many bioactive substances found in human bodies. The self-assembly of peptides has piqued considerable interest with groundbreaking advancements achieved in experimental research. However, it is still a big challenge to establish comprehensive theoretical model to accurately describe the behavior of peptide self-assembly. Current peptide self-assembly designs primarily rely on experimental outcomes and general rules, which is inefficient and susceptible to human errors. In recent years, thanks to rapid advancements in computer techniques and theoretical methods, computational research has become a vital tool in complementing experimental research with rapid development witted in this field. This review delves into the description of peptide self-assembly, covering relevant sequences, structures, morphologies, rules, and application areas. It places particular emphasis on the recent progress in computational methods such as molecular dynamics (MD) simulations and machine learning (ML) techniques in the study. Finally, we provide a perspective on the application of computational methods to expedite exploration in the realm of multi-peptide self-assembly.