|
Abstract:
|
Integrating multi-omics data has been meaningful in gaining a better understanding of disease development. Early childhood caries (ECC), a common and serious health condition, occurs when bacteria on the teeth break down compounds, altering the acidity in the mouth. Therefore, it makes sense to jointly analyze the metabolome and microbiome in dental plaque to investigate the complex associations that lead to tooth decay. Currently, simple correlation-based analyses are typically used to discover links between metabolites and microbiota. However, such methods are not well-suited for these data since they do not accommodate the excessive zeros existing in both the metabolome variable and the microbial variable. Here, we explore joint bivariate negative binomial models that account for the excessive zeros in both metabolome and microbiome data to reveal true correlations between them and compare correlation patterns within pathways between the ECC group and the normal group. We take into consideration the prior known pathways (e.g., in KEGG) to uncover how microbiome and metabolome together are associated with the development of ECC in an interpretable way.
|
