Integrating Metagenomic and Bayesian Analyses to Evaluate the Performance and Confidence of CrAssphage as an Indicator for Tracking Human Sewage Contamination in China
Haiyang Chen, Chang Liu, Yuezhao Li, Yanguo Teng
Abstract
Recently, crAssphage has been proposed as a human-specific marker for tracking fecal contamination. However, its performance has always been validated in a limited number of host samples, which may obscure our understanding of its utility. Furthermore, few studies have quantified confidence of fecal contamination when using crAssphage. Here, we evaluate the performance and confidence of crAssphage by analyzing a large panel of metagenomic data sets combined with Bayesian analyses. Results demonstrate that crAssphage exhibits superior performance with high host sensitivity and specificity, indicating its suitability for tracking human fecal sources. With the marker, a high confidence (>90%) can be obtained and particularly, multiple samples with positive results provide a near certainty of confidence. The application of crAssphage in the sediments of three Chinese urban rivers shows a high confidence of >97% of human fecal contamination, suggesting the serious challenge of sewage pollution in these environments. Additionally, significant correlation is observed between crAssphage and antibiotic resistance genes (ARGs), expanding the utilization of crAssphage for pollution management of ARGs. This study highlights the benefit of using metagenomic-based analysis for evaluating the performance and confidence of microbial source tracking markers in the coming era of big data with increasing resources in available metagenomic data.