Table 2

Number (prevalence) of bimodal genes as determined by compounding model selection criteria in the two different groups of microarrays.

(17,881 genes analyzed)

Criteria for Bimodal Assessment

Group A

Group B


p-val < 0.001a

430 (2.40%)

40 (0.224%)


+ misclassification area <0.1b

408 (2.28%)

36 (0.201%)


+ >10% points in each componentc

33 (0.185%)

33 (0.185%)


+ gender p-val for heterogeneity >0.05d

19 (0.106%)

21 (0.117%)


Columns represent the two groups of microarrays from which the bimodal genes were determined: Group A - 71 microarray chips all from the same lot, time span, and fluidics station; Group B - 47 microarray chips from two consecutive lots and from the same time span and fluidics station. The final row represents the total number of interesting bimodal genes for muscle expression in this population as estimated by the various chip groupings. A total of 17,881 genes were analyzed for potential bimodality.

a p-value from a likelihood ratio test with 6 d.f.

b calculated as proportion of misclassified area to unity

c points in each component tallied as number to the left and right of the two components' intersection

d from assessment in Group A

Mason et al. BMC Genomics 2011 12:98   doi:10.1186/1471-2164-12-98

Open Data