Biotechnological advances in torularhodin production: artificial neural networks as a tool for improving and biocompatibility studies.

Torularhodin is a bioactive carotenoid synthesized by certain microorganisms through complex cellular processes regulated by factors like nutrient availability. However, enhancing torularhodin production is a challenging task that requires costly and time-intensive experimental approaches. To address these limitations, computational modeling and simulation have become valuable tools for predicting and optimizing carotenoid biosynthesis. Among these techniques, polynomial models derived from multiple regressions provide useful insights but often struggle with the nonlinear nature of biological systems. In contrast, Artificial Neural Networks (ANNs) offer a more flexible alternative, improving predictive accuracy where traditional models fall short. This study aimed to optimize torularhodin production in Rhodotorula glutinis using ANN-based simulations and Response Surface Methodology (RSM) while also assessing the biocompatibility of the crude extract containing carotenoids. An experimental design with two independent variables (Tween 80 and malt extract) was implemented to evaluate their impact on torularhodin yield. ANN modeling successfully increased torularhodin production by approximately 10.69%, demonstrating its efficiency in bioprocess optimization. Additionally, microbial biomass extracts containing carotenoids exhibited biocompatibility in the Chorioallantoic Membrane assay, suggesting potential applications in pharmaceutical and food industries. These findings reinforce the importance of ANN modeling in optimizing microbial carotenoid production for sustainable biotechnology.