I was wondering what is the best approach to normalising Chip-seq samples using the deeptools software.

In our data set we're using an antibody to pull down specific regions of the genome. But when I do the coverage plot I already get very low coverage over the whole genome. This make sense, as i have only a small portion of the genome mapped. This made me also thinking about the normalisation method I should use with these samples.

If I normalise them based on the effective genome size, don't I decrease the heights of the peaks, as only a very small portion of this genome is truly mapped?

Wouldn't the RPKM or CPM (BPM) would be better in this case?

thanks