Gene expression and adaptive noncoding changes during human evolution

BMC Genomics

Table 2 Functional correlations between overall differential expression and noncoding adaptation

Collection #	sets	r_r	p(r_r)	q(r_r)
C3:TFT (transcription factor targets)	472	0.35	< 0.0001	< 0.0005
C2:CGP (chemical and genetic perturbations)	645	0.28	< 0.0001	< 0.0005
C7:All (immunologic signatures)	1723	0.14	< 0.0001	< 0.0005
C4:CGN (cancer gene neighborhoods)	88	0.5	0.012	0.045
C6:All (oncogenic signatures)	109	0.22	0.019	0.057
C4:CM (cancer modules)	87	0.29	0.036	0.09
C5:CC (GO cellular components)	37	0.33	0.14	0.27
C3:MIR (microRNA targets)	109	0.11	0.14	0.27
C1:All (positional gene sets)	9	0.17	0.34	0.53
H:All (hallmark gene sets)	25	0.089	0.35	0.53
C2:CP:Reactome (Reactome gene sets)	39	0.058	0.4	0.55
C2:CP:KEGG (KEGG gene sets)	11	−0.17	0.7	0.88
C2:CP:Other (other canonical pathways)	6	−0.49	0.8	0.93
C5:BP (GO biological processes)	123	−0.28	0.95	0.96
C5:MF (GO molecular functions)	44	−0.41	0.96	0.96

All MSigDB collections we considered are listed except for C2:CP:BioCarta, in which no gene set contained at least 50 genes analyzed in a previous study of adaptive noncoding changes. # sets is the number of gene sets we considered in the collection. r _r is the rank correlation between rank-biserial correlations for enrichment with differential expression (each of which is a rank-biserial correlation as in Table 1) and rank-biserial correlations for enrichment with noncoding adaptation (each of which is likewise a rank-biserial correlation). p(r _r) is the one-tailed p-value of r _r via permutation test with 10,000 permutations; collections are ordered by increasing p(r _r), which reflects r _r, # sets, and patterns of overlap among gene sets in the collection

ISSN: 1471-2164