Can I run the spiker.py to call peaks using control sample without spike-in read?

[petitmingchang]

I have two files, an Input.bam for control and an H3K27ac.bam for calling peaks. My case is that only the H3K27ac.bam contains spike-in reads but Input.bam does not. I already used split_bam.py to separate reads for H3K27ac.bam. Can I still use spiker.py to call peaks? If yes, what scaling factor should I set for the control file? Thank you so much for the help!

[liguowang]

In this scenario, I do NOT recommend you use spiker.py to call peaks. You probably need to redo the sequencing of the Input sample with spike-in control.

1. If you have spike-in for both ChIP and input, use this tool to do the peak calling. Spiker.py will scale the sequencing depth according to external control.
2. If you don't have spike-in controls for both ChIP and Input, then use MACS2 to do the peak calling. In this case, MACS2 will scale the sequencing depths of ChIP and Input to the same level, please note, this step assumes the starting amounts of chromatin are the same (which in general cannot hold).

[petitmingchang]

@liguowang I see, thank you so much for your clear explanation. I have three pairs of ChIP data from H3K4me1, H3K4me3, and H3K27ac between two conditions and would like to compare those peaks between conditions. Since I didn't have the spike-in control, I had used MACS2 to call peaks for each of ChIP data. However, I wonder do we still have the chance to do the normalization according to our spike-in data?

[liguowang]

yes, if you think there is global increase/decrease histone methylation levels, you have to have external controls. You can at least manually scale the data at least for differential analyses.

[petitmingchang]

@liguowang Got it, thank you very much for your prompt reply!