engleski

Computational methods for using known antimicrobial peptide sequences to identify or design novel non-homologous peptides

We shall describe two approaches for rapidly finding novel antibiotic peptide classes from animals also in view of the loss of such natural antibiotics due to accelerated species extinction. Both approaches are also designed to breach homology bottleneck when looking for promising lead peptide candidates for the development of peptide antibiotics. The first approach deals with identifying elusive selectivity properties of antimicrobial peptides as quantified by measured therapeutic index (TI). Introduced sequence moments has helped us to construct algorithms not only for predicting a TI from the primary structure in the case of natural amphibian peptides, but also for designing peptide antibiotics with optimized TI, non-homologous to any known natural or synthetic antimicrobial peptides. For one class of such antibiotics, the adepantins [1], we shall compare the theoretical predictions with experimental activity and selectivity results. The second approach allowed us to identify numerous novel putative antimicrobial peptides in the EST database, by applying our observation that, as a rule, specific subclasses of conserved signal sequences exist that are exclusively associated with antimicrobial peptides. As it turns out, at the present time this observation has been most useful for identifying novel amphibian and teleost antimicrobial peptides. We shall discuss these results, their implications, and experimental verifications. [1] D. Juretić, D. Vukičević, N. Ilić, N. Antcheva, A.Tossi, (2009) J Chem Inf Model 49, 2873.

antimicrobial peptides; design; computational

nije evidentirano