Physico-chemical features for recognition of antimicrobial peptides

Abstract

Concerns over antibacterial resistance have antimicrobial peptides (AMPs) garnering attention as potential targets for new antibacterial drugs [1]. Wet-lab development of AMP-based drugs hinge on understanding the relationship between AMP sequence and activity [1]. In support of such efforts, we devise a method to highlight position-based physico-chemical features related to activity. We do so in a focused analysis of the mature peptide fragments of cathelicidins; a populous sequence-diverse family of well-studied a-helical AMPs [1]. We employ features based on the AAIndex [2], an extensive collection of documented physico-chemical amino acid properties, and Support Vector Machine (SVM) to recognize cathelicidins from a set of carefully designed decoy sequences. Our results demonstrate that these features are very useful in elucidating specific residue positions and properties related to AMP activity.