FOCuS: a metaheuristic algorithm for computing knockouts from genome-scale models for strain optimization

Although handful tools are available for constraint-based flux analysis to generate knockout strains, most
of these are either based on bilevel-MIP or its modifications. However, metaheuristic approaches that
are known for their flexibility and scalability have been less studied. Moreover, in the existing tools,
sectioning of search space to find optimal knocks has not been considered. Herein, a novel computational
procedure, termed as FOCuS (Flower-pOllination coupled Clonal Selection algorithm), was developed to
find the optimal reaction knockouts from a metabolic network to maximize the production of specific
metabolites. FOCuS derives its benefits from nature-inspired flower pollination algorithm and artificial
immune system-inspired clonal selection algorithm to converge to an optimal solution. To evaluate the
performance of FOCuS, reported results obtained from both MIP and other metaheuristic-based tools
were compared in selected case studies. The results demonstrated the robustness of FOCuS irrespective
of the size of metabolic network and number of knockouts. Moreover, sectioning of search space coupled
with pooling of priority reactions based on their contribution to objective function for generating smaller
search space significantly reduced the computational time.