How the global level of histone markers are changed by treatment

Question

[Farzaneh]

Hi everyone, I recently received a Chip seq dataset to analyze. A cell line has been under a treatment and the antibody they used was H3k9me3 and H3k27me3, I already called the peaks and annotated them, but I don't know how should I answer their main question of:

How the global level of histone markers are changed by treatment.

Is there any specific region I should visualize? How about for showing the genome wide changes?

I tried to find stuff out there to help me out, but was not successful.

Thanks,

Answer 1

[ATpoint]

I would interpret global as an attempt to see the big picture rather than individual regions. In this case I would perform a differential analysis to identify the regions that have significantly different read counts between conditions. As both histone marks are broad peaks I would probably try the window-based approach proposed in the csaw. There is an entire book about the package which is an awesome read, see here. Such an analysis would give you a list of regions that change, which could be visualized using MA-or Volcano plots. Interpretation whether the changes are "global" are then on you. If you have like 10k regions total, and 20 change, then this is not global. If 30% change then one could probably make a point that the treatment has notable influence on histone modification remodeling. What is the experimental design, so how many experimental replicates (if any) per condition?

[Farzaneh]

Thanks for reply and suggestions. How do you consider that different read counts? What is the threshold for that? I called the peaks with SICER2, I have read counts and some other scores like p-value, and even something called peak score.

It's on a cell line, 3 conditions (control and two different treatments). Just one input per this cell line, they don't have input for each treatment [which I know is not usual]. They have three replicates per condition. So, I also have a bit of concern about how to merge these peaks from different replicates.

[ATpoint]

Please go through the linked book. It answers all these questions.