Back

Kinetic modeling of continuous meta-fermentation quantifies metabolic activity in a complex microbial system

Koga, T.; Kajimoto, K.; Ishizu, M.; Hamada, H.; Miyamoto, H.; Oshiro, M.; Sakai, K.; Tashiro, Y.

2025-11-14 bioinformatics
10.1101/2025.11.13.688382 bioRxiv
Show abstract

One of the serious drawbacks of a complex microbial system is the difficulty in quantifying the metabolic activity of each microorganism. A kinetic model of predominant microbial species was constructed for continuous meta-fermentation in a complex microbial system at several dilution rates (D). The introduction of biomass and lactic acid inhibition terms improved model accuracies at D = 0.05 h-1 and 0.4 h-1, respectively. The coefficient of determination (R2) and root mean square error (RMSE) improved from 0.577 and 5.21 to 0.972 and 0.759, respectively, with the inhibition term at D = 0.05 h-1. The inhibition terms resulted in good R{superscript 2} (0.996) and RMSE (1.27) values at D = 0.4 h-1. By solving the Michaelis-Menten equation in the constructed models, the species flux (SF) was calculated to estimate the metabolic activity (formation and consumption) of each microorganism. At D = 0.05 h-1, Caldibacillus hisashii contributed to lactic acid production at 0.333 g/L/h, whereas Clostridium cochlearium consumed lactic acid at 0.203 g/L/h, suggesting cross-feeding of a metabolite. It is therefore possible to account for consumption that cannot be considered in gene-derived calculations, indicating that this is a promising analytical method for investigating the dynamic behavior of complex microbial systems. Kinetic models for continuous meta-fermentation at several D values were developed. A new metabolic analysis method is proposed to estimate the activity of microbial species in a complex microbial system.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.