Figure 2

Bound and regulated genes are more conserved in predicted binding potential than control genes of comparable binding potential. (A) Predicted binding potential for the 600 bp upstream of every protein-coding gene in S. cerevisiae was calculated using GOMER and a Leu3 position weight matrix, and the calculations were repeated for the 600 bp upstream of all orthologous genes from six other species (Methods). The red circle shows the ROC AUC value obtained from comparing the 9 S. cerevisiae genes that were identified as bound and regulated in low activity experiments with all other genes in the same genome; the black circles are the values calculated for orthologous genes in six related species. The red box plot show the distribution of ROC AUC values obtained from 20 sets of 9 control sequences that were selected to closely approximate the predicted binding potential of the bound and regulated genes (Methods). The black box plots are comparable but are calculated using the orthologs of the genes chosen from S. cerevisiae; the open circles are the box plot outliers. For both experimental and control genes, if an ortholog was undefined in a genome, that gene was simply omitted from the analysis for that genome. Note that the predicted binding potential of bound and regulated genes is maintained in the distantly related species S. castelli and S. kluyveri to a much greater extent than control sequences. (B) Same as panel A except the genes defined as bound and regulated required high Leu3 activity for their detection in either the ChIP or expression assays, or both. There are 45 such genes, 31 of which required high Leu3 activity for detection in both assays. The remainder were detected at low activity in one of the assays (Fig. 3).