MiRKAT-MC

MiRKAT-MC performs microbiome kernel association testing to detect associations between microbiome composition and multi-categorical (nominal and ordinal) outcomes across independent and clustered datasets.

Key Features:

• Versatility with Multi-Categorical Outcomes: Supports analysis of nominal and ordinal multi-categorical outcomes, including disease subtypes and ordered severity levels.
• Handling Clustered Data: Accommodates clustered study designs such as family-based and longitudinal studies to account for within-cluster correlation.
• Incorporation of Ecological Distances: Integrates multiple ecological distance measures into the association testing framework to capture varied patterns of microbiome composition.
• Statistical Rigor with Pseudo-Permutation Strategy: Employs a computationally efficient pseudo-permutation strategy that maintains nominal type I error rates and improves statistical power, as demonstrated in simulations.

Scientific Applications:

• Clinical microbiome studies: Applied to studies where multi-categorical outcomes (nominal or ordinal) are central to investigating microbiome-disease relationships.
• Complex study designs: Used for analyses of family-based and longitudinal datasets to identify associations while accounting for clustering.
• Discovery of phenotype associations: Enables detection of associations between microbiome composition and disease phenotypes that may inform understanding of disease mechanisms and therapeutic targets.

Methodology:

Performs kernel-based association testing integrating multiple ecological distance measures and employs a computationally efficient pseudo-permutation strategy; performance was evaluated using comprehensive simulations and supports analysis of nominal and ordinal outcomes in independent and clustered datasets.