I am performing allele specific ChIP-Seq analysis. First I split allele specific reads using SNP positions. This I did both for Input and ChIP sample. Then I used MACS2 for peak calling as follows:
macs2 callpeak -t Sample_rep1_allele1.bed -c Input_rep1_allele1.bed -B -f BED -g mm -n Sample_rep1_allele1
macs2 callpeak -t Sample_rep1_allele2.bed -c Input_rep1_allele2.bed -B -f BED -g mm -n Sample_rep1_allele2
macs2 callpeak -t Sample_rep2_allele1.bed -c Input_rep2_allele1.bed -B -f BED -g mm -n Sample_rep2_allele1
macs2 callpeak -t Sample_rep2_allele2.bed -c Input_rep2_allele2.bed -B -f BED -g mm -n Sample_rep2_allele2
#Merge both the replicates
Later, I tried to look for monoallelic peaks (peaks unique to one of the allele). I used bedtools to get the peaks which are not shared between the two alleles:
bedtools intersect -wa -wb -a Sample_rep1-rep2-allele1_peaks.bed -b Sample_rep1-rep2-allele2_peaks.bed -v > Sample_allele1_specific_peaks.bed
bedtools intersect -wa -wb -b Sample_rep1-rep2-allele1_peaks.bed -a Sample_rep1-rep2-allele2_peaks.bed -v > Sample_allele2_specific_peaks.bed
When I visualised some of the random peaks on UCSC browser by uploading bigwig file, I noticed some of the peaks which were unique to allele1 was also present in allele2 with almost equal height and not present in either of the bed files (which made me consider them as monoallelic).
At some of the region's it was completely unique and doesn't overlap means they were completely monoallelic.
I am not sure why they all were not unique (monoallelic) when I tried to extract peaks only specific to one of the allele using bedtools (-v).
I thought it might be possible in either of peak calling step (allele1 or allele2) that peak was not called as the peak, may be because of some internal MACS2 calculations or during normalisation with respective controls. However I am not sure for the exact issue and the solution for this.
I hope I am clear with my question.
It will be very helpful if someone can suggest me what I am doing wrong or missing something.