Motif analysis

For both mouse and human, motif discovery was performed using MEME 4.12.0 as part of the MEME-ChIP suite⁷⁰ (‘-nmotifs 3 -minw 15 -maxw 45 –dna -revcomp) on 400 bp centered on the summits of all ZFP30-bound regions. In mouse, the set containing the intersection of at least two regions, irrespective of the day, was used (in2). ROC curves were calculated similarly to ref. ⁴¹. Briefly, we considered a set of p, positive sequences of length l (ZFP30 peak sets, union) and a set of n negative sequences of length l (shifted ZFP30 sequences). All sequences were scored for the motif using Fimo (in MEME-ChIP suite) with the parameters --output-pthresh 0.1 and considering only the best-scoring motif hit/sequence, thus obtaining two vectors of p and n scores. The statistic w of a Wilcoxon-test between both vectors was computed and used to display the ROC curves in Supplementary Fig. ^2N and compute the AUC as w /(n x m). The fraction of peaks with motifs displayed in Supplementary Fig. ^2O corresponds to motif hits obtained using Fimo and the parameters—output -pthresh 0.001. To determine which TFs potentially co-bind with ZFP30, we expanded our motif analysis using HOMER findMotifs with the parameters mopt = “-len 6,8,10,12,14,16 -size 100”, reporting known motifs in Supplementary Data ⁵.