Identifying and Assessing Interesting Subgroups in a Heterogeneous Population

Figure 5

(a) value distribution of two-sample -statistics for detecting differentially expressed genes from a lung cancer data comparing relapse versus no relapse and (b) the corresponding false discovery rate (FDR) estimate. In (b), the dashed line is the standard FDR estimate and the solid line is from our proposed procedure. The -axis in (b) denotes the ranking of genes where higher ranking corresponds to higher statistical significance.

(a) Histogram of values

(b) FDR