Characterizing Genes by Marginal Expression Distribution - Advances in Computational Science and Engineering - page 170

NORMAL

Fig. 3. The components (one and two) of mixture models of normal distribution fit to
the marginal distribution of the CSN3 gene is shown

Table 3. Loglikelihood, D and p-value, averaged over each probe of the GDS592
dataset

Normal Lognormal Gamma

#Comp Loglik D p-value Loglik D p-value Loglik D p-value

1 -728.47 1.14e-3 0.99 -85.04 7.70e-5 0.99 -492.66 1.72e-3 0.99

2 -683.53 1.44e-3 0.99 -47.39 7.73e-5 0.99 -437.79 1.05e-3 0.99

3 -661.46 2.70e-3 0.99 -46.64 9.20e-5 0.99 -487.58 1.26e-3 0.99

4 -637.21 5.79e-3 0.97 -46.01 6.17e-5 0.99 -474.41 1.19e-3 0.99

5 -612.37 9.44e-2 0.88 -45.08 7.45e-5 0.99 -465.15 8.36e-4 0.99