what do negative values from bigWigAverageOverBed output mean?

Question

[AB]

Hi,

This might seem basic, but I have a bed file with promoter regions I am interested in and Bigwig files from my ChIP-Seq experiment after subtracting input from treatment files using deeptools bamcompare. I want to know the coverage of my bw files. I came across the bigWigAverageOverBed utility from UCSC that I used but I am not sure I understand the results. Can anyone help me understand how this utility works and what negative values of the score mean?

Thanks

Answer 1

[Devon Ryan]

If the normalized input level is, on average, higher than that of the ChIP (at least in the regions of interest) then the average value will be negative.

[AB]

Thank you Devon. I have data that looks like this

chr chr_start chr_end ENSEMBL Diff_cell undiff_cell

chr2 76542036 76825604 ENSMUSG00000051747 -3.09314 7.20316

-3.093 and 7.2 are average coverage values of the differentiated cell sample and undifferentiated cell sample. Now i need to know if this gene and other genes with similar values are enriched in any GO category. Would it be right to log transform the average coverage (form bigWigAverageOverBed), calculate the fold change and use the FC to create a ranked list of genes for GSEA ? In that case, how do i log transform large negative values ?

[Devon Ryan]

The only way for fold-changes to make sense is if you put a floor of 0 on the values and add a small value to everything.