Enrichment of specific sequences during the SELEX process. (A) Position weight matrices built around the most enriched sequence for four different TFs (see Methods for details). The height of the letter at each position is directly proportional to the incidence of the indicated base in sequences where all other bases exactly match the most enriched sequence. Note that clear enrichment of sequences is observed after one or two SELEX rounds, and that two separate experiments for TCF4 result in a very similar enrichment pattern. In the first cycle, the algorithm used here detects incorrect binding profile for PRDM1 (asterisk) due to a low number of the relatively long consensus sequences. The enrichment of high-affinity sequences can, however, be detected by seeding the algorithm with consensus from the later cycles (see Supplemental Fig. S4A). (B) The fraction of all fragments containing the most enriched sequence from the third SELEX cycle plotted as a function of the SELEX cycle.
