Skip to main content

Table 4 Experimental comparison of hypothesis testing methods. The microarray data consists of 1973 genes partitioned into 12 quasi independent n1 = 2 and n2 = 2 subsets using replicate experiments of 4 control and 4 treatment runs [26, 27]. The number of genes detected as differentially expressed using each method is listed for an α = 0.01 significance level (there was no correction for multiple testing) and different prior degrees of freedom (ν0). The average number of common genes for the 66 unique comparisons of the 12 subsets is also listed along with the % of commonality (common/detected) for each test.

From: A two-sample Bayesian t-test for microarray data

ν0 DetectedPE DetectedMP CommonPE CommonMP % CommonPE % CommonMP
0 0 46.2 0 6.4 0 13.8
2 3.3 72.5 0.2 21.0 6.3 29.0
4 29.7 88.0 8.7 30.9 29.2 35.1
8 61.5 95.2 23.4 37.9 38.1 39.8
16 83.3 103.9 34.0 42.6 40.1 40.1
400 107.0 108.1 51.2 51.5 47.9 47.6
1972 118.8 119.0 54.7 54.7 46.0 46.0
3944 109.1 109.2 48.6 48.7 44.6 44.6