|
Abstract:
|
In human microbiome studies, the sequencing reads data are often summarized as counts of bacterial taxa at various taxonomic levels represented as a taxonomic tree. In addition, multiple measurements of microbiome from the same subject are often obtained to assess the difference in microbial composition across body sites or time points. Existing models for such count data are often restricted in modeling the covariance structure of the counts and cannot handle paired multinomial data. We propose a new probability distribution for paired multinomial count data, which allows flexible covariance structure of the count data and can be used to model repeated measured multivariate counts. Based on this new distribution, we develop a statistic to test the difference in compositions based on paired multivariate count data. We demonstrate the application of the test for analysis of count data observed on a taxonomic tree in order to test difference in microbiome composition across body sites and to identify the subtrees with different subcompositions. Our simulation shows that the proposed test has correct type 1 errors and increased power compared to some commonly used methods.
|
