ChIP sequencing analysis

To explore key TF binding patterns throughout the genome and specifically at TAD boundaries, we obtained publicly available SP1, CTCF, and TCF4 ChIP-seq data for the HCT116 colorectal cancer cell line from the Gene Expression Omnibus using the accession numbers {"type":"entrez-geo","attrs":{"text":"GSM1010902","term_id":"1010902"}}GSM1010902 (SP1), {"type":"entrez-geo","attrs":{"text":"GSM1010903","term_id":"1010903"}}GSM1010903 (CTCF), and {"type":"entrez-geo","attrs":{"text":"GSM782123","term_id":"782123"}}GSM782123 (TCF4). Two replicates were available for SP1 and CTCF, hence replicate profiles were averaged together. Profiles were then binned at 100kb resolution to be compatible with our RNA-seq and Hi-C data. The percentage of nonzero CTCF and TCF4 peak intensities that coincided with TAD boundaries called on the Hi-C data are reported in Supplementary Tables 5 and 6. Additionally, because TFEA analysis revealed that SP1 is the most highly enriched TF near the CEACAM loci on chromosome 19, we evaluated SP1 peak intensities at these loci. To evaluate the influence of TCF4 binding on the CEACAM loci, we aligned the TCF4 ChIP-seq profile to the region as well. Because ChIP-seq data for SP1 and TCF4 were acquired from different experiments, direct com- parison of their binding behavior at the CEACAM loci was facilitated by performing min-max normalization on each profile.