how to get promoter coordinates?

Question

[epigene]

I'm curious how you get all the promoter coordinates? Specifically, how you deal with one genes having multiple isofroms and multiple promoters? If there are 30k genes, do you ensure 30k promoters, i.e. one promoter per gene? What's the best practice? There is no standard way of doing it from what I can tell. So how do you do it?

Answer 1

[Devon Ryan]

Personally I get the promoter region for each transcript and then possibly merge them within a gene if they overlap (assuming that what I'm doing next is compatible with merging them). BTW, this is really easy to do in R with GenomicFeatures and a TranscriptDb.

[epigene]

Thanks for the quick reply. That's actually what I was doing as well! But if I define 2k upstream of TSS to be a promoter, by doing the merging, you could end up with very long promoters (>10k), which is a bit of concern to me. What do you think about that?

[Devon Ryan]

Whether that makes sense or not is dependent on what you want to do downstream. Obviously if you need to do any plotting of enrichment as a function of proximity to the TSS then merging wouldn't be a good idea.