Microba Research Discovery Report

Differentially abundant microbial functions were identified using the univariate methods ANOVA or LMER (linear mixed effect regression) on clr transformed relative abundance data, Fisher's exact test, and Aldex2 (ANOVA-like Differential Expression). Aldex2 was run on read count data. Fisher's exact test was used to test for diffrerences in the presence and absence (detection rate) of microbial functions across study groups.

LMER is used for repeated measures data, using random effects to control for correlation between samples from the same subject. Fixed effects are included for treatment groups, time, and treatment over time, where appropriate. The LMER P values correspond to a nested model test of the significance of including the corresponding fixed effect.

ALDEx2 uses subsampling (Bayesian sampling) to estimate the underlying technical variation. For each subsample instance, center log-ratio transformed data is statistically compared across study groups and computed P values are corrected for multiple testing using the Benjamini–Hochberg procedure. The expected P value (mean P value) is reported, which are those that would likely have been observed if the same samples had been run multiple times. The expected values are reported for both the distribution of P values and for the distribution of Benjamini–Hochberg corrected values.