BayMeth: improved DNA methylation quantification for affinity capture sequencing data using a flexible Bayesian approach

Table 1 Performance assessment for IMR-90 analysis (chromosome 7)

SssI depth	Number of bins	Method	Mean bias	Mean of squared differences	Spearman correlation	Wald	Highest posterior density	Quantile
[ 0,4]	305,638	BayMeth	-0.04	0.08	0.36	0.74	0.89	0.89
		BayMeth (SssI-free)	-0.19	0.20	0.23	—	—	0.24
		Batman	0.22	0.14	0.31	—	—	0.43
		MEDIPS	-0.38	0.26	0.29	—	—	—
		BALM	-0.48	0.33	0.32	—	—	—
(4,7]	22,196	BayMeth	0.05	0.05	0.65	0.84	0.88	0.87
		BayMeth (SssI-free)	-0.01	0.08	0.42	—	—	0.68
		Batman	0.16	0.07	0.61	—	—	0.34
		MEDIPS	-0.23	0.11	0.45	—	—	—
		BALM	-0.27	0.15	0.60	—	—	—
(7,14]	28,871	BayMeth	0.06	0.04	0.69	0.84	0.86	0.86
		BayMeth (SssI-free)	0.02	0.05	0.57	—	—	0.79
		Batman	0.16	0.07	0.65	—	—	0.28
		MEDIPS	-0.21	0.10	0.49	—	—	—
		BALM	-0.21	0.11	0.66	—	—	—
(14,27]	28,928	BayMeth	0.05	0.03	0.76	0.81	0.85	0.82
		BayMeth (SssI-free)	0.08	0.04	0.72	—	—	0.70
		Batman	0.15	0.06	0.73	—	—	0.23
		MEDIPS	-0.20	0.09	0.59	—	—	—
		BALM	-0.15	0.07	0.75	—	—	—
(27,168]	28,719	BayMeth	0.02	0.03	0.79	0.73	0.86	0.78
		BayMeth (SssI-free)	0.11	0.04	0.77	—	—	0.48
		Batman	0.11	0.05	0.75	—	—	0.20
		MEDIPS	-0.22	0.10	0.67	—	—	—
		BALM	-0.14	0.06	0.76	—	—	—

Results are shown for bins with a truth depth larger than the 25% quantile (cutoff is 33 reads), stratified into five groups by SssI depth. Shown are the number of bins per group, mean bias, MSE, Spearman correlation and coverage probabilities at 95% level.

ISSN: 1474-760X