A Bayesian calibration model for combining different pre-processing methods in Affymetrix chips

BMC Bioinformatics

Table 5 Performance of simulated data on ϕ and λ

	E(λ_j\| y) [CI 95%]	E(ϕ_j\| y) [CI 95%]	sd (λ_j)	sd (ϕ_j)
Method 1 (λ = 1, ϕ = 1)	1.0 [1.0-1.0]	1.0 [1.0-1.0]	n.a.	n.a.
Method 2 (λ = 0.5, ϕ = 0.5)	0.5 [0.49–0.51]	0.5 [0.49–0.51]	0.03	0.04
Method 3 (λ = 0.5, ϕ = 2)	0.50 [0.49–0.51]	2.0 [1.97–2.04]	0.04	0.05
Method 4 (λ = 2, ϕ = 0.5)	2.0 [1.96–2.07]	0.5 [0.48–0.51]	0.06	0.04
Method 5 (λ = 2, ϕ = 2)	2.0 [1.97–2.04]	2 [1.99–2.03]	0.05	0.03

The table reports the posterior mean with the 95% credibility interval for the measurement error parameters, together with the true values set in the simulation. For all the parameters, the posterior mean coincides with the true value and is characterised by a small variability around it. The values are averaged over 10 runs. The empirical standard deviation over the 10 runs is also reported in the table. Note that the stardard deviation is not available for Method 1 as it is assumed to be the reference (λ = 1, ϕ = 1).

ISSN: 1471-2105