Hello,

I`m currently working with ChIP-seq data related to a specific epigenetic modification (H3K4me3), under different development stages of the organism that I'm working with (a fungus), and with different strains (genotypes) of it. I would like to produce PCA (Principal Component Analysis) plots from those ChIP-seq data. Hence, I would like to ask:

• Which tool(s) should I utilize in R to produce PCA plots?
• What kind of files should I use as an input, for my ChIP-seq data, in order to produce a PCA plot?

Relevant info: I have done part of the analysis by using the packages below:

Bowtie2: for the mapping and to generate BAM files. MACS2 for the peak calling for the ChIP-seq. bamCoverage: to generate bigwig files for visualization.

Any help will be highly appreciated,

Best wishes,

Manolis

Not specific for the tagged package, but generally:

PCA is performed on a matrix of counts, typically normalized for library size and composition (e.g. by edgeR or DESeq2), and typically log2-transformed. The matrix has peaks/regions as rows and samples as columns. You would need to call peaks first and then produce a (consensus) peak list, and use that to create a count matrix, e.g. with featureCounts. In R that could be directly done with the Rsubread package.

PCA, if you want a single function, could be done by DESeq2::plotPCA() but many other packages can do that as well. Or you produce a PCA manually with prcomp() and then do the plotting of results downstream. That is not specific to Bioconductor though, so you might want to follow a guided tutorial.

Hi ATpoint,

Many thanks for your explanation, and for your suggestions as well.