Structure-activity trends analysis between amino acid compositions and minimal inhibitory concentrations of antimicrobial peptides

Antimicrobial peptides (AMPs) provide large structural libraries of molecules with high variability of constitutional amino acids (AAs). Highlighting structural organization and structure-activity trends in such molecular systems provides key information on structural associations and functional conditions that could usefully help for drug design. This work presents link analyses between minimal inhibitory concentration (MIC) and different types of constitutional AAs of anti-Pseudomonas aeruginosa AMPs. This scope was based on a dataset of 328 published molecules. Regulation levels of AAs in AMPs were statistically ordinated by correspondence analysis helping for classification of the 328 AMPs into nine structurally homogeneous peptide clusters (PCs 1-9) characterized by high/low relative occurrences of different AAs. Within each PC, negative trends between MIC and AAs were highlighted by iterated multiple linear regression models built by bootstrap processes (bagging). MIC decrease was linked to different AAs that varied with PCs: alcohol-type AAs (Thr, Ser) in Cys-rich and low Arg PCs (PCs 1-3); basic AAs (Lys, Arg) in Pro-rich and low Val PCs (PCs 4-8); Trp (heterocyclic AA) in Arg-rich PCs (PCs 6, 7, 9). Aliphatic AAs (more particularly Gly) showed MIC reduction effects in different PCs essentially under interactive forms.