A multi-objective hybrid machine learning approach-based optimization for enhanced biomass and bioactive phycobiliproteins production in Nostoc sp. CCC-403

Abstract

The cyanobacterial phycobiliproteins (PBPs) are an important natural colorant for nutraceutical industries. Here, a multi-objective hybrid machine learning-based optimization approach was used for enhanced cell biomass and PBPs production simultaneously in Nostoc sp. CCC-403. A central composite design (CCD) was employed to design an experimental setup for four input parameters, including three BG-11 medium components and pH. We achieved a 61.76% increase in total PBPs production and an almost 90% increase in cell biomass by our prediction model. We also established a test genome-scale metabolic network (GSMN) for Nostoc sp. and identified potential metabolic fluxes contributing to PBPs enhanced production. This study highlights the advantage of the hybrid machine learning approach and GSMN to achieve optimization for more than one objective and serves as the foundation for future efforts to convert cyanobacteria as an economically viable source for biofuels and natural products.

Keywords: Genetic algorithm; Machine learning; Nostoc sp.; Phycobiliproteins; genome-scale metabolic network (GSMN).