How to normalize ChIPseq and RNAseq. Especially for bam file.

Question

[feng920308]

Recently, I have read "Gcn4 Binding in Coding Regions Can Activate Internal and Canonical 5 0 Promoters in Yeast". However, I did not understand how they did the normalization for the ChIPseq and RNAseq.

For ChIPseq "Gcn4 occupancy profiles were obtained from the alignment (.bam) files using the bioinformatics toolbox from MATLAB. To allow the comparison between different samples, each profile was normalized such that the average occupancy for each chromosome was equal to one."

For RNAseq "RNA reads are normalized by combining all reads on both DNA strands for genes on the same chromosome and setting the average read density per nucleotide to one."

Because they did the strand-specific RNAseq so they need to combine all read on both DNA strands for genes.

Can anyone help me?

Answer 1

[ATpoint]

This method section is missing essential details on how exactly they normalized, so only for read depth or also for composition, and with which tools. I'd simply ignore it and do a proper normalization with established tools. I am not saying what they did was improper, just that the method section is not at all sufficient to reproduce it. Unfortunately a very common problem with method sections.

I have a post here on normalization on ATAC-seq (but the same applies for ChIP-seq): A: ATAC-seq sample normalization (quantil normalization) RNA-seq is basically the same. If you need browser tracks then follow this code and feed the size factors into a tool such as deeptools to scale the bigwigs. It all starts from a matrix of raw counts. You do not normalize bam files but the files you produce from it, which are the aforementioned count matrices (e.g. for differential analysis) or bigwigs for IGV visualization or other applications such as profile plots.

[feng920308]

Thank you very much!