Hi,
I'm using DeepTools to normalize chip experiments and I'm grateful to this wonderful application. I have read the following issue and know the importance of signal-to-noise normalization besides Per-million normalization. Normalize BigWig's for # of reads in peaks and sequencing depth
I have the following samples.
sample1_chip.bam sample1_input.bam
sample2_chip.bam sample2_input.bam
First I use bamCompare to obtain
sample1_chip-input.bw sample2_chip-input.bw
which are first normalized between each chip and input sample. Next I want to normalize between sample1 and sample2 .
I know there is bigwigCompare to normalize 2 bigwig files, but the ouput gives only 1 bigwig file.
My aim is to obtain each normalized
Normalized_sample1_chip-input.bw Normalized_sample2_chip-input.bw
so that I can feed them to computematrix and compare the plot profile.
Is there anyway to do this?
Many thanks!
1 answer
Thanks for immediate response!
Yes, scaling factor calculated from edgeR seems working. I first transformed .bw to .bedgraph and scaled using this amount, after putting it back to .bw file. Also thanks for making me clarify the following idea.
Correcting for inputs is difficult and from what I understand still not really resolved towards a gold-standard method because ChIP and input samples are globally different towards its distribution. Inputs are commonly used during peak calling, but that's it.
Best regards,
Log in to answer this question.
Why don't you simply apply the steps in the linked thread to obtain two normalized bigwig files? This should correct for library size and library composition. Correcting for inputs is difficult and from what I understand still not really resolved towards a gold-standard method because ChIP and input samples are globally different towards its distribution. Inputs are commonly used during peak calling, but that's it. If you use TMM to calculate size factors based on the peaks then you also normalize for signal/noise ratio, wouldn't that be siffucient?