By Vila?a, P.; Maia, P.; Rocha, I.; Rocha, M.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
The identification of a set of genetic manipulations that result in a microbial strain with improved production capabilities of a metabolite with industrial interest is a big challenge in Metabolic Engineering. Evolutionary Algorithms and Simulated Annealing have been used in this task to identify sets of reaction deletions, towards the maximization of a desired objective function. To simulate the cell phenotype for each mutant strain, the Flux Balance Analysis approach is used, assuming organisms have maximized their growth along evolution. In this work, transcriptional information is added to the models using gene-reaction rules. The aim is to find the (near-)optimal set of gene knockouts necessary to reach a given productivity goal. The results obtained are compared with the ones reached using the deletion of reactions, showing that we obtain solutions with similar quality levels and number of knockouts, but biologically more feasible. Indeed, we show that several of the previous solutions are not viable using the provided rules.