How to show statistical significance of overlap of CTCF binding sites

Question

[samuel]

I have carried out RNA-Seq for lncRNAs. I have 8 lncRNAs that are differentially expressed and all 8 overlap CTCF binding sites when viewed in Ensembl genome browser. How do I go about proving that this is statistically significant and not happening by chance? I have read in papers where they have done something similar but they never state how they did it. Also could someone tell me where I actually find out exactly how many CTCF binding sites there are? (for human) . I cannot find this information from Ensembl. Many thanks.

Answer 1

[rpolicastro]

You can calculate this using a hypergeometric test, which is also commonly used for GO enrichment. In short, it will calculate the probability of randomly picking 8 CTCF sites without replacement, and having all 8 sites match to your 8 target transcripts.

All of the CTCF binding sites can be accessed from the Ensembl regulation dataset in biomart here.