Much less peaks compared to other replicates

Question

[zau saa]

Hi! I'm doing ATAC-seq data analysis now. There're 3 replicates in one condition. However, I got much less peaks in a replicate under a certain condition. Compared to other replicates in this condition, only 1719 peaks were called in this replicate (more than 30k peaks were called in other two replicates). I'm confused how to deal with this replicate. Shall I discard it in further analysis?

Raw reads were trimmed by Trimmomatic and then aligned to hg38 by bowtie2 with parameter --very-sensitive and -X 2000. Reads mapped to chr1-22 and XY were preseverd. PCR duplicates were removed by Picard. Peaks were called by MACS2 with parameter -f BAMPE