Congratulations to our colleague Kristaps Berzins, who has successfully defended his Master's thesis titled “Application of artificial intelligence in stoichiometric model-based selection and metabolic engineering of organism strains for biotechnology” at the Faculty of Computing, University of Latvia.
The analysis of genome-scale stoichiometric models is becoming an increasingly popular method for analysing organisms, searching for potential modifications, and generally helps to get rid of unnecessary laboratory experiments. At the same time, the size of metabolic reactions, metabolites, and gene databases is significantly increasing. Both reasons lead to the need for methods and algorithms that can process ever-increasing data and facilitate the creation and processing of metabolic models.
Kristaps developed a kStrainAlgorithm (see figure 1), which uses genetic algorithms to perform three main functions: 1) identify the necessary reactions for the target metabolite production of the genome-scale metabolic model, 2) adds a set of reactions that improve fluxes of the target metabolite and 3) gets rid of reactions that have little or no effect on the fluxes of the target metabolite. Thus, ideas for potential modifications of the organism or possible laboratory experiments with these organisms have been obtained.
Figure 1. Algorithm flowchart.
The algorithm successfully fulfils the tasks given to it and the recommendations for modifications of the genome-scale model are obtained in a satisfactory time, according to which the strains of organisms and their modifications can be selected. Algorithm parameters and functions can be easily modified and adapted to different situations. Thus, the kStrainAlgorithm has the potential to solve not only the problems of reactions and metabolite addition but also other problems of analysing genome-scale models.