How to decide H3K27me3 peaks in promoter region

Question

[izzy.yichao.cai]

Hi all!

I have a H3K27me3 bed file that is predicted by SICER and H3K4me3 narrowPeak file predicted by MACS2.

I want to know what are those monovalent promoter(that is with only one mark in the promoter region)

This is what I did:

1.I use bedtools to find the H3K4me3 summits that has no overlap with H3K27me3 broad peaks(bedtools intersect -v)

2.Then I find the overlap between step 1 result and the slopped annotation bed file that have the coordinate from -500 bp and +500 bp of TSS of genes

3.Similar to step 1 & 2, I try to find the promoter that only marked by H3K27me3 broad peaks

Here is what I got: The result of monovalent promoter of H3K4me3 seems good and reasonable, while the result of H3K27me3 is not quit good. I got many results that H3K27me3 peaks spread across gene body but not in the promoter region. (I check various transcript and it seems that there are no H3K27me3 peaks that fall in promoter region).

Any hints about this?

Thanks a lot !

[Denise CS]

Assuming this is in human, I can see some regions annotated as having promoter activity (ENSR00002122957 and ENSR00002122958) in the 5' end of ABCA1 where there are H3K27me3 peaks. Same for ABO (ENSR00002128324). Apparently no H3K27me3 peaks in the 5'end of ADD2 (ENSR00001927961).

Answer 1

[Sinji]

I think I can help with this, but you'll have to walk me through what you want to do since i'm having trouble fully understanding the question.

1. You're looking for genes that contain H3K4me3, but no H3K27me3, in a -500/+500 bp region around the TSS?

2. You're then looking for genes that contain H3K27me3, but no H3K4me3, in a -500/+500 bp region around the TSS?

You should take into account that 'promoter' region is arbitrary and depending on the literature can range from -250/+250 to -3000/+3000 from the TSS of a gene.