Integration of taxa abundance and occurrence frequency to identify key gut bacteria correlated to clinics in Crohn's disease.

ABSTRACT

Bacteria abundance alternation in the feces or mucosa of Crohn's disease (CD) patients has long been applied to identify potential biomarkers for this disease, while the taxa occurrence frequency and their correlations with clinical traits were understudied. A total of 97 samples from the feces and gut mucosa were collected from CD patients and healthy controls (HCs), 16S rRNA-based analyses were performed to determine the changes in taxa abundance and occurrence frequency along CD and to correlate them with clinical traits. The results showed that bacteria communities were divergent between feces and mucosa, while the taxa abundance and occurrence frequency in both partitions showed similar exponential correlations. The decrease of specific fecal bacteria was much more effective in classifying the CD and HCs than that of the mucosal bacteria. Among them, Christensenellaceae_R-7_group and Ruminococcus were predicted as biomarkers by using random forest algorithm, which were persistently presented (> 71.40% in frequency) in the feces of the HCs with high abundance, whereas transiently presented in the feces (< 5.5% in frequency) and mucosa (< 18.18% in frequency) of CD patients with low abundance. Co-occurrence network analysis then identified them as hub taxa that drive the alternations of other bacteria and were positively correlated to the circuiting monocytes. The loss of specific bacteria in the healthy gut may cause great disturbance of gut microbiota, causing gut bacteria dysbiosis and correlated to immune disorders along CD, which might not only be developed as effective noninvasive biomarkers but also as therapy targets.