ChIP-seq signal across bed regions

Question

[SEJAL]

I have a bed file with genomic coordinates, I used bedtools multicov to get the reads mapped to each of the regions. I have to compare the signal across 2 samples. But my samples don't have replicates. Which normalisation and what tools should I be using?

Answer 1

[mark.ziemann]

If you only have two replicates, then you have limited options. Differential expression tools like edgeR and DESeq2 can be used for ChIP-seq, but only if you have two or more replicates. The reason for this is that the p-values from these tools are trying to quantify sources of variation including technical and biological noise. It just doesn't make sense when you don't have proper replicates.

That being said, there are other ways to analyse this data. One idea could be to use a peak caller like MACS2 which will identify the parts of the genome that are enriched in a sample as compared to a control. Another possibility is to use something like GFOLD which provides a foldchange statistic in the case of no-replicates. GFOLD provides p-values but these are not realistic because there is no real measure of within group variation.