A comparative analysis of mutual information methods for pairwise relationship detection in metagenomic data

BMC Bioinformatics

Table 2 AUCs for each metric under exploitative relationships with zero inflated counts

		Log normal	Exponential	Negative binomial	Gamma	Beta negative binomial
n = 50	MINE	0.888 ± 0.041	0.866 ± 0.040	0.590 ± 0.063	0.743 ± 0.057	0.620 ± 0.066
	NWJ	0.821 ± 0.051	0.784 ± 0.055	0.646 ± 0.062	0.744 ± 0.055	0.582 ± 0.067
	DOE	0.459 ± 0.070	0.530 ± 0.070	0.531 ± 0.064	0.479 ± 0.068	0.486 ± 0.066
	MIC	0.835 ± 0.046	0.817 ± 0.050	0.558 ± 0.068	0.660 ± 0.059	0.526 ± 0.068
	KSG (3)	0.648 ± 0.063	0.692 ± 0.060	0.572 ± 0.070	0.658 ± 0.062	0.563 ± 0.066
	LNC (7)	0.689 ± 0.063	0.668 ± 0.062	0.592 ± 0.064	0.668 ± 0.060	0.611 ± 0.066
	Partitioning (5)	0.754 ± 0.057	0.671 ± 0.063	0.556 ± 0.067	0.662 ± 0.061	0.580 ± 0.071
n = 200	MINE	0.991 ± 0.006	0.989 ± 0.007	0.754 ± 0.056	0.938 ± 0.023	0.810 ± 0.050
	NWJ	0.945 ± 0.027	0.961 ± 0.021	0.762 ± 0.058	0.908 ± 0.035	0.737 ± 0.060
	DOE	0.708 ± 0.060	0.646 ± 0.066	0.552 ± 0.067	0.540 ± 0.069	0.599 ± 0.065
	MIC	0.892 ± 0.038	0.920 ± 0.030	0.715 ± 0.058	0.805 ± 0.050	0.556 ± 0.066
	KSG (9)	0.881 ± 0.039	0.881 ± 0.039	0.713 ± 0.059	0.737 ± 0.059	0.688 ± 0.059
	LNC (7)	0.839 ± 0.044	0.902 ± 0.036	0.659 ± 0.063	0.770 ± 0.055	0.709 ± 0.062
	Partitioning (5)	0.853 ± 0.045	0.905 ± 0.033	0.714 ± 0.059	0.885 ± 0.038	0.682 ± 0.062

Best results for each prior distribution by relationship type are indicated by bolded font

ISSN: 1471-2105