Figure 4. Genome-Wide Scan of the Drosophila E1-E2 Model onto the Mouse Genome.

(A) Genome-wide rankings of circadian mouse genes in function of their sequence likelihood score.

The E1-E2 chained matrices are scanned along all mouse genes including ±2 kb flanking sequence (Methods). Only hits with average PhastCons scores above 0.5 are counted (in PhastCons 0 implies no conservation and 1 maximal conservation). The unconstrained result is shown in Figure S8. The name of the gene is always aligned to the right of its score line (in red), and its rank is indicated in brackets. A total of 7 of 13 expected circadian genes (Methods) are found above the 1% line (green dashed line), while we expected zero at this cutoff (p < 10⁻¹²). Notice Wee1 is just below the 1% line; E1-E2 is in the 3' region for this gene (Figure S7). Known circadian genes represented are Cry1, Cry2, Per1, Per2, Per3, Dbp, Tef, Hlf, Wee1, Bhlhb2 (Dec1), Bhlhb3 (Dec2), Nr1d1 (RevErbα), Nr1d2 (RevErbβ), Arntl (Bmal1), and Clock. The latter two are expressed in anti-phase with respect to known target and are mainly controlled by Ror orphan receptors and their repressors RevErbα,β [70]. Thus, Clock and Bmal1 are not included in the test set.

(B) Phase distribution of all conserved hits (as in [A]) with scores above 15 bits, and which show cycling in the liver in [54] (Fourier component F24 > 0.1; Methods).