Average Width Of H3K4Me3 Peaks

Question

[aditi.qamra]

Hi,

I used CCAT to call peaks for the H3k4me3 mark in my samples and the average peak width I noticed is ~3kb

I wanted to get a sense check on these nos. ? Usually in cancer samples, is that the sort of profile that is seen for H3k4me3. I tried looking this up in literature but I dint get enough data pts and the few that I did showed an average profile of 1.5kb.

Does anybody have any references on this ?

Thanks !

Answer 1

[Sukhi Singh]

From H3K4me3 in mESC (mouse embryonic stem cells), what I know is that the promoters which have the tri-methylated nuclesomes at active. This width, I plot as an average profile in and around +/-3KB of TSS, which is sufficient to answer. Its hard to say the exact width, but I would assume it to be a max of 3KB so it might be fine.

As an example, I took this screenshot from the cistrome webpage. Its obvious that the average width would be around 2-3KB but I am not sure if that be different in the cancer cell line.

[aditi.qamra]

Hi Sukhdeep.. Thanks for this.. This gives me some consolation. The peaks im looking at are infact active promoter regions and some of my peaks even go up to 5kb or 8kb. Makes me wonder if taking a union of overlapping peaks is a good idea ? Are promoter regions really that long or Im lumping them all and missing out on information..

[Sukhi Singh]

There might be different cases why the peak looks broad

1. it might be the promoter of an upstream gene
2. unannotated promoters
3. a small (noise) peak joining the tail of a normal peak making it look peak
4. If you center your peaks at TSS and make a union, it will take into consideration the variable widths of the peaks, as for most of the peaks, 3KB upstream would come clean and for some not, so with the union you will see that, highest in the centre and then a decreasing gradient.

Cheers

[tonja.r]

Do you know why H3K3me3 shows a small peak before the TSS and a more sharper peak after TSS?