We curated Position Frequency Matrices from cisBP representing a total of 15,389 human motifs and 14,367 mouse motifs. To filter motifs to a representative subset, we first categorized motifs as high, medium or low quality, as is provided in the cisBP database. We then grouped all 870 unique human or 850 unique mouse TF regulators represented in the database and assigned these regulators to their most representative TF motif(s). To do this, we first iterated through each TF regulator to find all motifs associated with that regulator from the high-quality motif list. For these associated high quality motifs, we first computed a similarity matrix using the Pearson correlation of the motifs. To calculate the Pearson correlation between pair-wise motifs, the shorter motif was padded with an equal distribution of A,C,G,T. Then the Pearson correlation was calculated at every possible offset, and the maximum correlation of all offset comparisons was recorded. To select a representative subset of motifs for each TF regulator, we first found the motif correlated with the most other motifs at R>0.9. Treating that motif and all of
