Table 2 Mean and standard deviation discrimination accuracy (AUC) of the different methods for different effect sizes
From: coda4microbiome: compositional data analysis for microbiome cross-sectional and longitudinal studies
Method | n1 = n2 = 50 | n1 = n2 = 100 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Effect size | 1.25 | 1.5 | 2 | 5 | 10 | 20 | 1.25 | 1.5 | 2 | 5 | 10 | 20 |
coda4microbiome | 0.763 (0.12) | 0.943 (0.1) | 0.988 (0.05) | 0.999 (0.01) | 1 (0) | 1 (0) | 0.776 (0.12) | 0.891 (0.08) | 0.987 (0.02) | 1 (0) | 1 (0) | 1 (0.03) |
selbal | 0.795 (0.12) | 0.912 (0.11) | 0.993 (0.07) | 1 (0) | 1 (0) | 1 (0) | 0.761 (0.11) | 0.924 (0.09) | 0.987 (0.01) | 1 (0.01) | 1 (0) | 1 (0) |
aldex2 | 0.59 (0.06) | 0.6 (0.16) | 0.946 (0.13) | 1 (0) | 1 (0) | 1 (0) | 0.629 (0.11) | 0.898 (0.16) | 0.982 (0.02) | 1 (0) | 1 (0) | 1 (0) |
ancombc | 0.573 (0.04) | 0.603 (0.06) | 0.963 (0.07) | 1 (0.01) | 1 (0) | 0.954 (0.1) | 0.569 (0.03) | 0.847 (0.13) | 0.971 (0.02) | 1 (0.01) | 1 (0) | 1 (0.01) |
DESeq2 | 0.56 (0.12) | 0.851 (0.15) | 0.973 (0.04) | 1 (0) | 1 (0.01) | 0.882 (0.09) | 0.672 (0.11) | 0.92 (0.07) | 0.989 (0.01) | 1 (0.01) | 1 (0) | 1 (0) |
edgeR | 0.581 (0.12) | 0.841 (0.14) | 0.974 (0.03) | 1 (0) | 1 (0.01) | 0.925 (0.11) | 0.692 (0.11) | 0.921 (0.07) | 0.988 (0.01) | 1 (0.01) | 1 (0) | 1 (0) |
metagenomeSeq | 0.573 (0.04) | 0.593 (0.14) | 0.962 (0.06) | 1 (0.01) | 0.997 (0.04) | 0.929 (0.07) | 0.591 (0.03) | 0.829 (0.12) | 0.982 (0.02) | 1 (0.01) | 1 (0.12) | 0.972 (0.13) |
LinDA | 0.583 (0.04) | 0.604 (0.08) | 0.7 (0.09) | 0.859 (0.06) | 0.907 (0.05) | 0.883 (0.1) | 0.575 (0.03) | 0.662 (0.07) | 0.729 (0.1) | 0.936 (0.07) | 0.976 (0.02) | 1 (0.01) |