Machine Learning Approach Reveals the Assembly of Activated Sludge Microbiome with Different Carbon Sources during Microcosm Startup

Abstract

Activated sludge (AS) microcosm experiments usually begin with inoculating a bioreactor with an AS mixed culture. During the bioreactor startup, AS communities undergo, to some extent, a distortion in their characteristics (e.g., loss of diversity). This work aimed to provide a predictive understanding of the dynamic changes in the community structure and diversity occurring during aerobic AS microcosm startups. AS microcosms were developed using three frequently used carbon sources: acetate (A), glucose (G), and starch (S), respectively. A mathematical modeling approach quantitatively determined that 1.7–2.4 times the solid retention time (SRT) was minimally required for the microcosm startups, during which substantial divergences in the community biomass and diversity (33–45% reduction in species richness and diversity) were observed. A machine learning modeling application using AS microbiome data could successfully (>95% accuracy) predict the assembly pattern of aerobic AS microcosm communities responsive to each carbon source. A feature importance analysis pinpointed specific taxa that were highly indicative of a microcosm feed source (A, G, or S) and significantly contributed for the ML-based predictive classification. The results of this study have important implications on the interpretation and validity of microcosm experiments using AS.

Insightful Analysis and Prediction of SCOD Component Variation in Low-Carbon/Nitrogen-Ratio Domestic Wastewater via Machine Learning

“…Zhao et al [40] utilized six kinds of machine learning models to estimate the RBCOD and SBCOD in municipal wastewater with an R 2 higher than 0.80 by inputting oxidation-reduction potential (ORP) data. Kim et al [41] established high-performance (>95% of accuracy) ML models to predict the influ-ence of different feeding carbon sources (acetate, glucose, and starch) on the microcosm communities of activated sludge. Therefore, machine learning has demonstrated excellent performance in past research; these algorithms help us to identify the complex relationship between input factors and output factors, and achieve high levels of accuracy and generalization.…”