Advances in computational and data-driven methodologies for accelerating antimicrobial peptide design and discovery

Abstract

Molecular simulations of antimicrobial peptides (AMPs) have advanced significantly in recent years, driven by improvements in theoretical models, simulation techniques, and computational resources. While early efforts were limited by small system sizes and short timescales, modern approaches enable simulations of large biomolecular assemblies over microsecond durations. Atomistic molecular dynamics (MD) simulations provide detailed insights into peptide-membrane interactions, including adsorption, insertion, and pore formation, while coarse-grained (CG) models facilitate exploration of larger systems and longer timescales, capturing collective behaviors such as peptide aggregation and membrane remodeling. Enhanced sampling methods, including umbrella sampling and replica exchange have improved the characterization of free energy landscapes associated with the AMP activity. The convergence of machine learning (ML) and molecular simulations has revolutionized antimicrobial peptide (AMP) discovery, transforming it from empirical screening to a rational, data-driven discipline. This integration has given rise to a new computational design paradigm that unites molecular dynamics simulations, surrogate ML modeling, and sequence–structure–activity principals for targeted peptide design. Persistent challenges remain in force-field precision, sampling depth, accurate representation of non-standard residues and post-translational modifications such as amidation and lipidation. Looking ahead, coupling ML-driven simulations with automated synthesis and real-time experimental feedback is poised to establish a closed-loop discovery framework. This synergy between computation, robotic automation, and artificial intelligence defines the next frontier of predictive and scalable AMP design and development.