Bayesian modeling of ChIP-chip data using latent variables

BMC Bioinformatics

Table 3 Computational results for the simulated datasets.

Method	Total	ND	FD	r	p-value
Bayesian Latent	50.5 (0.58)	2.3 (0.33)	2.8 (0.57)	0.9545 (0.0080)	--
tileHMM	48 (0.77)	4.2 (0.57)	2.2 (0.55)	0.9250 (0.0107)	0.02
BAC	2934.7 (6.60)	0 (0)	2884.7 (6.6)	0.0609 (0.0003)	0.00
Wilcox	56.1 (0.95)	3.9 (0.48)	6.4 (0.62)	0.9221 (0.0088)	0.007
t-scan	78.9 (2.11)	3.1 (0.31)	27.6 (1.71)	0.9047 (0.0089)	0.0003
EB t-Scan	71.5 (1.52)	3.0 (0.39)	20.9 (1.38)	0.9176 (0.0068)	0.001

"Total" denotes the average number of bound regions identified for each of the 10 datasets, ND denotes the number of true bound regions that are not discovered by the algorithm, FD denotes the number of false bound regions discovered by the algorithm, r is the adjusted Rand index, the number in the parentheses is the standard error, and "EB t-scan" refers to the empirical Bayesian t-scan method proposed by Ji and Wong [6].

ISSN: 1471-2105