Comprehensive comparison of twenty structural characterization scales applied as QSAM of antimicrobial dodecapeptides derived from Bac2A against P. aeruginosa.

Informative numerical characterizations of amino acid residues are essential for quantitative sequence-activity modeling (QSAM). To date, a variety of structural characterization methods based on local amino acids have been proposed. However, limited detailed reports are available using same datasets and modeling methods to compare the ability to characterize structures of amino acids. Here, we evaluate the characterization capability of 20 descriptor sets on a set of antimicrobial peptides (AMPs) derived from Bac2A against P. aeruginosa. Results display the models by FASGAI, z-scales, VHSE, DPPS, HESH and ProtFP descriptors present qualified predictive capability. Moreover, the structural characterization of the studied AMPs should involve the hydrophobic, bulky and electronic properties of amino acids; besides, the secondary structural information should not be ignored. In parallel, the FASGAI-based model exhibits a more robust prediction than other models, and reasonably describe the structure-activity relationship of the studied dodecapeptides, which is in line with the reported experimental observations. This work provides references for methods of structural characterization as applied in QSAM of AMPs against P. aeruginosa.