Development of a rapid antimicrobial screening method for natural products using genetic programming

The importance of natural products in modern drug discovery is undisputed [1], not least thanks to the enormous structural diversity of plant secondary metabolites. However, the identification of compounds with a desired activity within this vast abundance remains a challenge. High-throughput screening is a successful approach, where simple tests gauge in a short time a compound's potential for further development. Our group employs screening techniques to search for novel antimicrobial agents from natural sources. In its simplest (and most straightforward) way, antimicrobial activity is detected by measuring the absorbance of a bacterial suspension after 24h incubation at a single wavelength. While this method is well-known and standardized [2], it is unspecific and fairly prone to assay interferences. With the goal to facilitate the screening process and thus increase its throughput, we have developed an assay that allows the antimicrobial activity screening of natural products against three microorganisms (Staphylococcus aureus, Escherichia coli, Candida albicans) simultaneously, as opposed to one at a time. In an attempt to get more comprehensive information, we mined the VIS and IR spectra of the bacterial/fungal suspension using genetic programming. Genetic programming is a machine learning algorithm, which follows the principal of biological evolution to develop (or evolve) a mathematical formula that, when applied to the spectra, yields a conjecture about present bacteria or fungi. This method has previously been used to successfully type single bacterial strains with the aid of IR spectroscopy [3]. Here, we present the development of this assay, with focus on the underlying mechanism, optimization process, the potential and pitfalls of genetic programming in bioactivity screening of natural products.

Keywords: Bioactivity, screening, high-throughput, Python, machinelearning, genetic programming, FTIR

References:

Newman DJ, Cragg GM. Natural products as sources of new drugs over the 30 years from 1981 to 2010. J Nat Prod. 2012; 75: 311 – 35.

Wikler M, Clinical and Laboratory Standards Institute. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically: approved standard. 8th ed. Wayne Pa.: Clinical and Laboratory Standards Institute; 2009.

Goodacre R. Explanatory analysis of spectroscopic data using machine learning of simple, interpretable rules. Vib Spectrosc. 2003; 32: 33 – 45.