Skip to main content
. 2020 Apr 21;36(14):4106–4115. doi: 10.1093/bioinformatics/btaa260

Table 2.

Results in analysis of the two real datasets

Testing the global hypothesis
Testing individual OTUs
Trait MiRKAT P-FL VE-freq VE-arcsin VE-omni VE-freq VE-arcsin VE-omni DESeq2
URT microbiome data
 Smoking 0.0019 0.0018 0.0070 0.0006 0.001 4 14 5 0
PPI microbiome data
 gPC1 0.22 0.19 0.13 0.47 0.19 0 0 0 14
 gPC2 0.36 0.19 0.19 0.19 0.26 0 0 0 49
 gPC3 0.24 0.31 0.30 0.21 0.29 0 0 0 29
 gPC4 0.16 0.088 0.013 0.080 0.021 0 1 0 13
 gPC5 0.0094 0.015 0.034 0.010 0.015 0 3 3 48
 gPC6 0.19 0.41 0.43 0.49 0.53 0 1 1 23
 gPC7 0.15 0.21 0.76 0.16 0.22 0 0 0 59
 gPC8 0.21 0.33 0.64 0.36 0.47 1 0 0 4
 gPC9 0.15 0.12 0.10 0.12 0.15 0 0 0 20

Note: P-FL, PERMANOVA-FL. For the global hypotheses, reported results are P-values. For the individual OTU tests, results reported are the number of OTUs detected at FDR = 10%. MiRKAT and PERMANOVA-FL results are based on the Bray–Curtis distance. In analysis of the URT microbiome data, MiRKAT and PERMANOVA-FL yielded P-values 0.005 and 0.0042, respectively, when the weighted-Unifrac distance was used, and 0.00078 and 0.0006 when the Hellinger distance was used.