We have the fly spike-in ChIP-seq data, I use the following tab to calculate the correlation factor,
File/Sample name DMSO_input Induced_Input DMSO_TF1 Induced_TF1 ChIP target TF1 TF1 Treatment DMSO induced DMSO induced Tags mapped to entire dm3 genome 82,362 84,401 2,179,901 662,991 Correction factor based on all dm3 alignments 1.024756562 0.304138124 Fly tags(unique) 82,362 84,401 2,179,901 662,991 Human tags(unique) 30,534,368 33,182,250 29,193,791 26,111,001 Scaled human tags (all dm3 alignments) 32,380,617 85,852,443 plusDox Global signal relative to Dox (all dm3) 1.060 2.941
Then I use macs2 bdgopt command to multiply the correlation factor for the treatment and control bdg file,
macs2 bdgopt -i TF1_treat_pileup.bdg -m multiply -p 2.941 -o TF1_treat_norm.bdg macs2 bdgopt -i TF1_control_lambda.bdg -m multiply -p 1.060 -o TF1_control_norm.bdg macs2 bdgcmp -t TF1_treat_norm.bdg -c TF1_control_norm.bdg -m ppois -o TF1_dm3_pvalue.bdg macs2 bdgpeakcall -i TF1_dm3_qvalue.bdg -c 7 -l 186 -o TF1_dm3_peaks_pvalue.bed
I am not sure whether I am right to do peakcalling like this, I just think if I increase the treatment signal 3 times, and the input signal is almost same with before, it will have a lot false peak calling. Can anyone please give me some suggestions how to do the peak calling after the dm3 normalization? Thanks.