Abstract
The growing antibiotic resistance issue requires the search for novel antimicrobial agents. Antimicrobial peptides (AMPs) are potent innate defence molecules that constitute useful templates for antimicrobial design. The establishment of databases facilitates the development of computer algorithms for peptide prediction. Most of the predictions are made based on the mature peptide sequences. These methods range from the simple rule-based to the more sophisticated machine learning. A shortcoming of these methods is that they all depend on the completeness of the representative templates used to train the programs. There are also, however, other methods that make predictions based on the conserved genes outside of the AMP gene box. The applications of these prediction methods to genomes and proteomes followed by experimental validation further accelerate the pace of peptide discovery. A combined use of genomic and proteomic approaches allows a more complete mapping of potential AMPs in a single organism. This chapter also discusses the major approaches for peptide design, including library screening, sequence shuffling, hybridization and de novo design. Database filtering technology has been developed and can be enhanced to design peptides with desired activity, structure and pharmaceutical parameters.