) show histograms of hypergeometric P
-values. In panel (a
), the P
-values are computed for 1000 random gene sets with respect to the yeast interaction data set of Duan et al.
). In panel (b
) the P
-values are computed with respect to a simulated data set for 250 random sets of 100 genes. In (c), the P
-values correspond to 174 gene sets regulated by a single transcription factor and studied in (11
), computed with respect to the yeast interaction data set of (3
). Panels (d
) are analogous to panels (a)–(c), but the P
-values are computed using the resampling approach. In each case, the resampling-based P
-values provide no evidence of colocalization of gene sets.