*4.2. Binding Site Predictions*

To identify the potential binding sites of the highly expressed TFs, we used a curated library of position weight matrices as previously described [28], only retaining those with high information content (≥10). Using the PRISM phylofootprinting method [28], we predicted binding sites for the highly expressed TFs in previously defined e9.5-specific enhancers, downloaded from the Gene Expression Omnibus (GSE65807) [13]. We filtered the predictions using the parameters as described in [102], only keeping significant (*p*-value ≤ 0.05) predictions with a match threshold of at least 0.8 that are conserved in the human genome (hg19).

Motifs with a similarity threshold of 0.8 were grouped, and only the motif with the highest number of occurrences was kept for further analysis. We determined the number of e7.5 and e9.5 enhancers with each motif and then calculated the fold by determining the proportion of enhancers containing a particular motif at e9.5 divided by the proportion at e7.5. Significance was determined using the hypergeometric test with Bonferroni correction for multiple comparisons.
