Combining RNA seq and H3K27ac ChIP seq data

Question

[thomasterTW]

I have 2 lists of comparisons between 3 conditions: (1->2 and 2->3) - List of differential gene expression (genes) (RNA seq data analyzed with DESeq2) - List of differential H3K27ac binding (peaks) (ChIP seq data analyzed with MAnorm)

I want to find correlation between gene expression dynamics and H3K27ac binding dynamics. For example a gene that is upregulated between condition 1 and condition 2, and also the H3K27ac binding near this gene goes up.

The problem here is that the H3K27ac differential binding data consists of peaks. Each peak has it's own p-value (significance of differential binding). When I annotate these peaks to the nearest gene (bedtools closest), most genes will have more than 1 peak in their surrounding, and also the statistical data (p-value) will be gone. How do I combine these different peaks which annotate to the same gene while retaining the statistical data? What I need is a list of genes ranked on significance (or fold change) of differential H3K27ac binding. When I have this I can compare this with the list of genes ranked on differential expression and find correlation.

Help is much appreciated.

[Mike]

You can also try with BETA tool. Binding and Expression Target Analysis (BETA) is a software package that integrates ChIP-seq of transcription factors or chromatin regulators with differential gene expression data to infer direct target genes. http://cistrome.org/BETA/

[thomasterTW]

I find this one very interesting, though I'm not sure how to use it. It says it will combine ChIP-seq of TF or chromatin regulators with cuffdiff results. However it seems that all functions in this software are based on transcription factors and their target detection. In their tutorials and readme they never mention chromatin regulators, only transcription factors. I want to simple get a list of upregulated genes between two conditions, where H3K27ac binding as well goes up. Can this be done with this software?

[David Gortler]

I found this article on finding patterns of transcription factors useful even though its slight;ly dated. https://link.springer.com/article/10.1186/gb-2012-13-3-r16

Answer 1

[ivivek_ngs]

You can do that, when you find differential acetylation sites remember, that H3K27ac are usually associated with enhancers. So you can always annotate the sites which are up and down between your condition of ChIP-Seq and retrieve the significant sites and finally annotate the enhancers from those region based on distance based to the nearest gene and then find the genes that correlate with the differentially expressed genes (DEGs if they are summarized into gene ids) this way you are selecting differential enhancers between your conditions that are significant and trying to correlate them with your DEGs to build the hypothesis that differentially marked enhancers are actually having an effect on your gene expression. Largely everything boils down to your conditions that you are working upon and try to see in literature if your phenotype is associated with enhancer marks or not. If so then the hypothesis can be established.

Alternatively you can always find peaks from your samples of condition and either used distanced based enhancer marking or HMM based like that of HistoneHMM or chromHMM, annotate for enhancers and try to see for those significant regions upon translating to nearest gene effect of those region in the gene expression or whether directly they capture the DEGs or not. Hope this helps. You can also take a look at this link

Take a look at this paper as well.

Answer 2

[Ming Tommy Tang]

Not exactly you want, but have a look for this tool http://cistrome.org/MARGE/ from Sheirly Liu's lab