Tracking Major Sources of Water Contamination Using Machine Learning

Abstract

Current microbial source tracking techniques that rely on grab samples analyzed by individual endpoint assays are inadequate to explain microbial sources across space and time. Modeling and predicting host sources of microbial contamination could add a useful tool for watershed management. In this study, we tested and evaluated machine learning models to predict the major sources of microbial contamination in a watershed. We examined the relationship between microbial sources, land cover, weather, and hydrologic variables in a watershed in Northern California, United States. Six models, including K-nearest neighbors (KNN), Naïve Bayes, Support vector machine (SVM), simple neural network (NN), Random Forest, and XGBoost, were built to predict major microbial sources using land cover, weather and hydrologic variables. The results showed that these models successfully predicted microbial sources classified into two categories (human and non-human), with the average accuracy ranging from 69% (Naïve Bayes) to 88% (XGBoost). The area under curve (AUC) of the receiver operating characteristic (ROC) illustrated XGBoost had the best performance (average AUC = 0.88), followed by Random Forest (average AUC = 0.84), and KNN (average AUC = 0.74). The importance index obtained from Random Forest indicated that precipitation and temperature were the two most important factors to predict the dominant microbial source. These results suggest that machine learning models, particularly XGBoost, can predict the dominant sources of microbial contamination based on the relationship of microbial contaminants with daily weather and land cover, providing a powerful tool to understand microbial sources in water.

Seasonal prevalence of bacteria in the outflow of two full-scale municipal wastewater treatment plants

“…Mathematical models and bioinformatics tools 38 , e.g. using machine learning 60 , are needed to better understand the phenomenon of seasonal variation and the importance of microbial uptake in water bodies in the future. The research confirmed that data from the outflow can be a valuable source of information on the operation of wastewater treatment plants, but in order to create such models, more data needs to be collected from both activated sludge and the outflow, which has not been studied so far.…”

Ontology-aware deep learning enables ultrafast and interpretable source tracking among sub-million microbial community samples from hundreds of niches

“…We further evaluated ONN4MST for samples from the same biome but with different characteristics. Previous studies have reported that microbial community samples from the soil with different characteristics possessed high diversity [ 31 ]. To evaluate the capability of ONN4MST for predicting samples from the same biome with different characteristics, we introduced another cohort, about the seasonal changes of the Hadza people’s gut microbial communities [ 26 ].…”

DeepToA: an ensemble deep-learning approach to predicting the theater of activity of a microbiome

“…Here we introduce DeepToA, an ensemble deep learning framework that aims at predicting the "theater of activity" (ToA) of a microbiome from the taxonomic and functional profiles of its metagenome. To the best of our knowledge, this is one of the first deep-learning approaches to focus on metagenomic data, rather than 16S community profile data, and to utilize both taxonomic and functional profiles (Shenhav et al, 2019;Zha et al, 2020;Wu et al, 2021).…”

“…Several deep-learning approaches have been developed to address microbiome-related questions, such as disease prediction (Oh and Zhang, 2020;Sharma and Xu, 2021), the annotation of antibiotic resistance genes (ARGs) (Li et al, 2021), microbial source tracking (Shenhav et al, 2019;Wu et al, 2021), and microbial community prediction (Thompson et al, 2019;Zha et al, 2020).…”

“…SourcePredict (Borry, 2019) uses dimension reduction followed by a KNN algorithm to classify and predict the origin of metagenomics samples. In a recent study (Wu et al, 2021), several machine learning algorithms are used to predict the dominant source of microbial contamination by using environmental and geographical data. The ONN4MST method uses an ONN (ontologyaware neural network) to embed biome ontology information into a hierarchical structure so as to improve performance of community-based microbial source tracking (Zha et al, 2020).…”