Improved quality metrics for association and reproducibility in chromatin accessibility data using mutual information

Table 2 Mean area under the curve across simulations

Statistic	Mean (95% CI)	Mean (95% CI)—Co-zeros removed	p-value^a	\(\sigma ^2\)^b
Top–down correlation	0.6881 (0.6860–0.6906)	0.6872 (0.6850–0.6895)	0.635	–
Pearson R	0.8284 (0.8237–0.8335)	0.6965 (0.6946–0.6984)	0.0	0.0201
R²	0.7026 (0.6951–0.7102)	0.5346 (0.5324–0.5368)	0.0	0.0329
Spearman \(\rho\)	0.9140 (0.9118–0.9162)	0.6686 (0.6665–0.6705)	0.0	0.0136
Kendall \(\tau\)	0.9096 (0.9074–0.9120)	0.6673 (0.6654–0.6691)	0.0	–
Kendall W	0.9570 (0.9559–0.9581)	0.8343 (0.8333–0.8353)	0.0	–
Normalized Mutual Information	0.8197 (0.8153–0.8241)	0.5055 (0.5045–0.5065)	0.0	0.016

^aThe p-value represents the test of differences in mean AUC after removal of co-zeros
^bVariation values were calculated during analysis of data from true ATAC-seq experiments

ISSN: 1471-2105