GC bias correction for Chip-SEQ

Question

[Farron16]

Hello,

I have a problem, being a little new on this field, I need to know why is the GC correction used ? is it linked to CPG islands ? And what does it do to correct the GC % in a chip SEQ analysis ?? If you have any answers I'll be grateful Regards

[colin.kern]

The way current sequencing technology works introduces biases for GC-rich areas of the genome, so they tend to get more reads than other areas. This is hard to account for in ChIP-seq because the areas of interest (promoters, enhancers, etc.) often are GC-rich themselves, so the bias is expected biologically and can be hard to discern from technical bias. I'm not sure if there are standard ways of adjusting for GC bias in ChIP-seq data.

Answer 1

[ATpoint]

As colin.kern says both the PCR amplification during library preparation as well as sequencing suffer from biases as a function of GC content in the genomic fragments. Still, by best knowledge there is neither a consensus that one should routinely correct for this, nor is there a well-tested and accepted pipeline to do so. It is (again by best knowledge) not standard to do it at all. There is a paper from Benjamini & Speed that addressed the problem of GC bias and suggest a correction method for DNA-seq, but I do not think it has widely been adopted. For RNA-seq there are approaches such as alpine from Michael Love (Bioconductor package and now implemented in software such as salmon), but for DNA-seq I would probably not bother with it by now. Not because it is not necessary, but because there are little or no standards and you would probably need to do some extensive confirmations that your correction is necessary, beneficial and does not introduce in silico biases.