Computer-aided design, structural dynamics analysis, and in vitro susceptibility test of antibacterial peptides incorporating unnatural amino acids against microbial infections

•An integrated protocol is described to design antibacterial peptides containing unnatural amino acids.•A variety of machine learning predictors are developed, optimized, and validated through a synthetic approach.•The predictors are used to guide computer-aided peptide design.•Four designed peptides are measured in vitro to have potent antibacterial activity.•The potent peptides can spontaneously embed into an artificial lipid bilayer.