MiRKAT

Test for overall association between microbiome composition data and phenotypes via phylogenetic kernels. The phenotype can be univariate continuous or binary (Zhao et al. (2015) doi:10.1016/j.ajhg.2015.04.003), survival outcomes (Plantinga et al. (2017) doi:10.1186/s40168-017-0239-9), multivariate (Zhan et al. (2017) doi:10.1002/gepi.22030) and structured phenotypes (Zhan et al. (2017) doi:10.1111/biom.12684). The package can also use robust regression (unpublished work) and integrated quantile regression (Wang et al. (2021) doi:10.1093/bioinformatics/btab668). In each case, the microbiome community effect is modeled nonparametrically through a kernel function, which can incorporate phylogenetic tree information.