ChIP-Seq Intensity Analysis

Question

[solveforj]

Hello,

I am wondering how I can measure ChIP-Seq intensity at any point of an aligned genome in windows of differing length (e.g. 1023-2000 bps and 1023 to 3004 bps). I am aware of R packages like csaw, which measure intensity at bins of a predefined length, but these windows will probably overlap with the specific windows I am targeting. Any suggestions?

Update - Here is some more detail about what I am doing. I am performing a peak calling on ChIP-Seq data for 1 protein. Then, I will look at the intensities for other proteins in the same regions as the peaks for the first protein. Can I use bedtools or deeptools for this specific application?

Thank you,
J.G.

[Devon Ryan]

Comparing signals near peaks across samples is easy enough with multiBigwigSummary or multiBamSummary from deepTools, but I'm not entirely sure what you want with differing sized windows. If by this you mean predefined regions of variable size then that's not a problem.

[solveforj]

Thank you for the advice. I have deepTools, and I believe I can use multiBamSummary in BED-file mode to analyze ranges of interest in ChIP-Seq data. How would I generate a BED file with these ranges?

Thank you,

J.G.

[venu]

@solveforj, I have successfully used deeptools for similar purpose. I was using the term re-quantifiation for this analysis.

Briefly, you can do this,

1. Generate normalized bigwig files from bam with bamCoverage function
2. Call peaks using MACS on the data for which you want to generate actual peak regions (and convert them to BED format using simple linux awk)
3. Use bed from step-2 to get normalized values of other chip-seq samples using multiBigWigSummary
4. Convert results from step-3 to a table and use ComplexHeatmap package from Bioconductor to do clustering

This analysis should give you an idea of how similar different samples are to the one you are testing (i.e from which you took the peak/bed regions).

[solveforj]

@venu, thanks so much! This helps a lot. Thank you for the advice.

• J.G.

[venu]

Providing few more details would help to answer. Like, why do you want to measure intensity in bins, for differential analysis or just to check comparative signal in different bins/regions. You can use bedtools, deeptools..etc to calculate intensity/read count and normalize the values.

[solveforj]

I will not be performing differential analysis on 2 or more genomes; just 1 genome. I am performing a peak calling on ChIP-Seq data for 1 protein. Then, I will look at the intensities for other proteins in the same regions as the peaks for the first protein. Can I use bedtools or deeptools for this specific application?

Thank you,

J.G.