Skip to main content
. 2022 Aug 19;10:130. doi: 10.1186/s40168-022-01320-0

Table 2.

Differential abundant analysis methods evaluated in this study

Method Addressing compositional effects Handling zeros Model Covariate/confounder adjustment
GMPR + Wilcox GMPR None Wilcoxon rank-sum test
TSS + Wilcox Total sum scaling (TSS)
Rarefy + Wilcox Rarefaction (TSS equivalent)
GMPR + DESeq2 Geometric mean of pairwise ratios (GMPR) Model (overdispersion) Negative binomial model
GMPR + edgeR GMPR Model (overdispersion) Negative binomial model
Wrench + MSeq Wrench Model (zero inflation) Zero-inflated log-normal model
RAIDA Reference Model (zero inflation) Zero-inflated log-normal model
ANCOM-BC Bias correction Pseudo-count Log-linear model
DACOMP Reference None Wilcoxon rank-sum test
LDM TSS None Linear model
Omnibus GMPR Model (zero inflation) Zero-inflated negative binomial model
Aldex2(Wilcox) Centered log-ratio transformation (CLR) Bayes Wilcoxon rank-sum test
Aldex2(glm) Generalized linear model (GLM)
GMPR + glm GMPR Model (overdispersion) GLM (quasi-Poisson)
Corncob TSS Model (overdispersion) Beta-binomial model
MaAsLin2 TSS Pseudo-count Log-linear model
eBay(Wilcox) CLR Empirical Bayes Wilcoxon rank-sum test
ZicoSeq Reference Empirical Bayes Linear model