RioNorm2

"RioNorm2" is a computational framework for the differential abundance analysis of sparse high-dimensional marker gene microbiome data. Recognizing the clinical significance of identifying differentially abundant microbial species about diseases, RioNorm2 addresses the unique challenges posed by microbiome data, including zero-inflation, over-dispersion, and variability in species' abundances across different scales.

Key Features and Functionalities:

Network-Based Normalization Technique: RioNorm2 introduces a network-based normalization method that identifies a group of relatively invariant microbiome species across samples and conditions. This innovative approach constructs size factors, enabling more accurate normalization of microbiome data.

- Two-Stage Zero-Inflated Mixture Count Regression Model: At the core of RioNorm2 is a two-stage zero-inflated mixture count regression model that enhances the differential abundance analysis. This model accounts for under-sampling and over-dispersion by categorizing microbiome species into two groups and modeling them separately, thereby increasing the method's flexibility and accuracy.

- Biological Insight Discovery: By applying RioNorm2 to a published dataset of metastatic melanoma, the framework has successfully uncovered biologically meaningful insights, showcasing its potential to contribute significantly to our understanding of the microbiome's role in diseases.