Table 1.
False positive rate and power for testing H0: β1 = 0 based on ZIGMMs and ZIBR for significance level at 0.05 for various sample sizes.
Table 2.
Parameter ranges in simulation studies.
Fig 1.
Empirical powers in four simulation settings under low effect scenario.
Fig 2.
False positive rates in all four simulation settings.
Table 3.
False positive rate and power for testing H0: β1 = 0 and H0: β3 = 0 from setting E for significance level at 0.05 for various sample sizes.
Table 4.
Proportions of significant taxa detected in four models with LMMs, NBMMs and ZIGMMs.
Fig 3.
The analyses of ZIGMMs(log), NBMMs and LMMs: minus log transformed p-values for the significant differentially abundant taxa at the 5% significance threshold between pregnancy and non-pregnancy groups for host factor effect (left panel) and interaction effect (right panel) from Model C.
Table 5.
Proportions of significant taxa detected in four models with LMMs and ZIGMMs.