Ensembl config option: what's the differences between ChIP-seq peak and ChIP-seq signal peak ?

Question

[michaelchen33]

Hi, everyone

When I use Ensembl to visualize Cmyc binding site in different cell types by opening both peaks and signal track style, I am curious why Cmyc ChIP-seq peak is not matching the ChIP-seq signal peak in HeLa-S3? They are apart from each other about 112-113 bp.

Please see the HeLa-S3 cell type as I mentioned above.

http://asia.ensembl.org/Homo_sapiens/Share/3443f4939f6685a4d7d6b8d04a3a5de93237444

Thanks

[Bert Overduin]

Tip: To send people a link to an Ensembl page without having to tell them how to configure the page, use the 'Share this page' option.

[michaelchen33]

Thanks a lot, I change the link.

[Emily]

That's odd, I had already replied to this question on Ensembl helpdesk over an hour before Bert suggested that you email Ensembl helpdesk. Please do not cross-post.

[michaelchen33]

Sorry about that, I did read the SWEMBL beginner's manual but I still don't understand the differences between ChIP-seq peak and ChIP-seq signal peak (Ensembl config option: peaks/signal track style), why is that their peak site are not the same.

[Emily]

The signal just shows the number of reads at any given position. The SWEMBL peak calling normalises this out in various ways, including correcting against the input sample (to remove experimental bias) and extending all the reads out to the full fragment length. This can alter where the peak of the peak actually lies.

[michaelchen33]

So does this means that the peak is more reliable than the signal peak (after adjustment), and that is why in some cell types we can still visualize signal peaks but there is actually no TF binding ?

[Emily]

Yes, the signal is pretty raw, the peak is it normalised so is better data.

[michaelchen33]

Got it. Thanks