I have a datasets where the coverage of input is relatively. Actually many reads were filtered out during deduplication step. When I performed peak calling, though peaks are called but low input coverage put me in doubt as if they are reliable or not.
One way I filtered out the peaks with low read count after counting. But since my interest is heterozygosity inside input reads the counting does not help. I also tried to call heterozygous bases from Input but then it is a round-about way to deal.
Is there any other way I can control number of reads in input to consider for peak calling?
Or can I merge both replicates of input to use as merged input (so I use same input for replicates of ChIP)?
Or we need more sequencing done for input to increase coverage (matter of cost)?
I am using MACS1/MACS2 for peak calling.
I would appreciate any suggestions.