Looking for the best way to annotate histone modicitation marks to genes.

Question

[thomasterTW]

I want to include differential binding of H3K27ac (ChIP-Seq) near genes in my differential gene expression (RNA-seq) analysis. The aim of this analysis is to see the expression dynamics in a cellline between 3 time points (proliferating (day 0) and differentiation (day 1.5 and 3)). I want to see for which genes the expression pattern correlates with the H3K27ac binding near these genes. For example gene X is not expressed at day 0 but is highly expressed at day 1.5 and 3. I want to know if there is a strong H3K27ac signal near this gene at day 1.5 and 3 and not at day0.

For the differential gene expression I used DESeq2 and for the ChIP-Seq I was planning to do the same (using DEseq to find genes with differential H3K27ac binding), and then later check for genes with correlating patterns. The thing is that for the RNA seq I used STAR to create the readcount files. I can create the readcount files for H3K27ac with htseq-count but I'm doubting if this is the right way to do this (comparing readcounts obtained with 2 different methods).

Would this be the right way to do this? And if not, could anyone give some advice on how to do it right, maybe there are better tools to do this or something that can do this all at once. I'm a beginning researcher without that much experience in this sort of analysis so any input would be highly appreciated.

Thanks in advance.

Answer 1

[SP]

I am also doing similar things. I would recommend to call peaks for H3K27ac first (it is a very abundant mark). You can annotate these peaks with nearest gene and where they belong on genome (e.g. TSS, intron, exon) Homer can be used for annotation. One straight approach would be to find which differential genes have H3K27ac peaks in the surrounding and then concentrate on these peaks for difference between condition. Second, look at specific peaks (e.g. only TSS or Intergenic peaks for DE gene) H3K27ac marks enhancer region much more strongly than TSS.

---

To compare H3K27ac gain or loss between condition. You can take the tag count for the peaks and see if it is changing within condition. I hope you have replicates for each conditions. Once you know the peaks you can just take their coordinates and do some heatmaps, box plots for tagcounts or use some differential peak callers This paper is comparing tools for differential peaks callers.
I hope it helps.