ProgPerm: Progressive permutation for a dynamic representation of the robustness of microbiome discoveries

BMC Bioinformatics

Table 2 Comparisons of identification performance among different methods.

Data	nsv	Method	FP	FN	Sensitivity	Specificity	Accuracy	RC
Set 1	70	WilPerm	1	3	0.96	0.97	0.96	0.80
		DESPerm	3.5	33.6	0.52	0.88	0.63	0.50
		DESeq	2	0	1	0.93	0.98	0.55
		LEfSe	19	2	0.97	0.37	0.79	0.16
		Logistic	1	4	0.94	0.97	0.95	0.58
	30	WilPerm	3	6.3	0.79	0.96	0.91	0.004
		DESPerm	2.1	12.8	0.57	0.97	0.85	0.094
		DESeq	4	0	1	0.94	0.96	0.17
		LEfSe	10	5	0.83	0.86	0.85	0.24
		Logistic	1	5	0.83	0.99	0.94	0.07
Set 2	70	WilPerm	6.3	0	1	0.79	0.94	0.02
		DESPerm	30	6.3	0.91	0	0.64	0.79
		DESeq	30	9	0.87	0	0.61	0.81
		LEfSe	30	4	0.94	0	0.66	− 0.04
		Logistic	5	40	0.43	0.83	0.55	− 0.88
	30	WilPerm	7	0	1	0.9	0.93	0.35
		DESPerm	19	0	1	0.73	0.81	0.80
		DESeq	3	7	0.77	0.96	0.90	0.80
		LEfSe	70	5	0.83	0	0.25	0.08
		Logistic	7	5	0.83	0.9	0.88	− 0.79

“WilPerm” stands for progressive permutation equipped with Wilcoxon tests. “DESPerm” stands for progressive permutation equipped with DESeq method. FP denotes number of false positives. FN denotes number of false negatives. Sensitivity measures the proportion of positives that are correctly identified. Specificity measures the proportion of negatives that are correctly identified. Accuracy measures the proportion of true positives and true negatives. RC denotes the rank correlation (Spearman’s \(\rho\)) between the true and estimated ranks of the features. Set 1 denotes the simulation data that varies the zero inflation parameter for each variables. Set 2 denotes the simulation data that varies the mean difference parameter for each variables

ISSN: 1471-2105