Differentially activated Enhancers with just H3K27Ac

Question

[lucatucciarone]

Hello Guys, Recently I got my H3K4me1 NGS data. Unfortunately seems like the ChiP-experiment didn't work out (as peaks are super-low and similar to the Input 😭).

I'm currently trying to understand how to, eventually, overpass the H3K4me1 as my H3K27Ac sequencing are very good in both CTR and Treated. I'm working in Mouse, in a pretty specific cellular population (similar but not definable as "Fibroblast") and all I have is H3K27Ac and Nuclear accessibility (NA-seq) DATA.

I was desperately trying to get some "Enhancers annotation" but it seem to be cell specific (to a certain degree). This is the site, if you are interested: http://www.enhanceratlas.org/index.php

Given that my cell is pretty "Niche" I was starting to think to simply integrating Differentially H3K27Ac Peaks covered with Na site 🤔.

Do you think is a viable strategy? Reading this article: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4517609/ I realized that Nuclear accessibility and H3K4me1 are both markers of eventually primed Enhancers. H3K27Ac is what should be essential to my goal, isn't it?

Thanks in advance

[Kevin Blighe]

I am not sure of how this relates to bioinformatics. You should seek advice from a colleague in your department.

[lucatucciarone]

hahaha it's more related to epigenetic in general, that's true 😊 Sorry for the question

[Kevin Blighe]

Non ti preoccupare

Answer 1

[jared.andrews07]

Yes, enhancers certainly vary by cell type/tissue. What is your ultimate goal? To find differentially active enhancers between your treated and control samples?

Regardless, your proposed strategy isn't without merit - I've seen several papers define enhancers in the same fashion, i.e. H3K27ac peaks overlapping DNase/ATAC peaks that don't overlap promoters/exons. Just be wary of trying to find differentially accessible regions in a quantitative manner - it's really hard to define the difference in a small DNase peak versus a "large" one. In general, I'd only do occupancy analysis for any accessibility assay so that you get binary peak/no peak calls, which will yield more believable differences between your sample groups. ChIP-seq is a different matter, and affinity-based analyses are a lot more valid for it as changes in levels of histone modifications can be experimentally validated (via dCas9-histone modifier fusions, etc).

[lucatucciarone]

Thank you very much for you answer! Yes! my goal is exactly to find differentially active enhancers between your treated and control samples. For this reason I'm assuming that the most important difference should be in H3K27Ac. My "strategy" would be to just integrate NA and H3K27Ac data for CTR and Treated (To have H3K27Ac peaks that a Nuclease accessibility site). subset this peaks to exclude the promoter and finally perform differential analysis on those "polished peaks".

I'm understanding that your main concern would be about the "differential NAs", isn't it? I don't think I need to integrate differential NAs and differential H3K27Ac peaks. I think I could answer my question by integrating the data as they are and then perform differential analysis. What do you think?

[jared.andrews07]

Yes, I just wanted to be sure you know the caveats of trying to look at different "levels" of accessibility. Otherwise, I think your proposed strategy is fine and will be unlikely to draw any ire from reviewers down the road.