|
Abstract:
|
A microbiome is the collection of all the micro-organisms, and their genes, in a defined environment. Such a community is inherently dynamic with the abundance of the microbes, as well as the relationships they form with one another changing with time. Longitudinal studies, in which data is collected repeatedly over time, are common in biomedical research. Such studies provide a way to investigate the temporal dependency between any pair of microbes. Copula models with mixed zero-beta margins have proven successful in identifying biologically meaningful microbial interactions from cross-sectional data, while accounting for the excessive zeros and compositional nature of metagenomic sequencing data. Our work aims to extend these methods to the longitudinal data setting. In particular, we specify a mixed margin copula model to detect temporally conserved microbial dependence. We illustrate our method using simulations and analysis of real microbiome data sets.
|
