Figure 4.

The fraction of conserved sequence and the identity of lineage-specific repeats (compared to their consensus sequence) in introns of genes expressed in different numbers of human tissues (means with LSD intervals). (A) fraction of conserved sequence in human introns. (B) identity of human Alu repeats. (C) identity of mouse B2 repeats. (D) fraction of conserved sequence in human introns, corrected simultaneously for identity of human Alu repeats, average between-species identity of conserved fraction and intron GC content, using the general linear model (GLM). (Only introns containing Alu repeats were taken in the latter case; the nonconserved regions were assumed to have zero identity.) For the effect of the number of tissues in the GLM, P <10–8. The picture was similar if the correction parameters were included in the model separately (one-by-one).

347fig4