HiCar-sample related query

[Adeel3Dgenomics]

I have been working on reproducing the GM12878 and H1 results from your original Hi-CAR paper, where you analyzed 4 and 3 replicates, respectively. However, I was only able to process 3 replicates for GM12878, as SRR15090019 did not process correctly. For H1, I successfully processed all 3 replicates.
After generating the interaction files at 5kb resolution (using default parameters), I observed the following:
GM12878 (3 replicates): 15,471 interactions
GM12878 (4 replicates, original): 48,517 interactions
Does the missing replicate (rep3) significantly impact the results, or could there be an issue in my processing pipeline?
H1 (3 replicates): 35,154 interactions
H1 (3 replicates, original): 34,400 interactions
This result is more consistent, suggesting the processing is largely correct.
I would appreciate any insights into whether missing rep3 in GM12878 could account for such a large difference in detected interactions. Also, could you remind me of any specific parameters you used in the original analysis that might be affecting the output?