Empty peak matrix

Hello,
Having issues running the peak calling pipeline.
Here is everything

import  scanpy as sc
import snapatac2 as snap
from MACS3.Signal.PeakDetect import PeakDetect
snap.__version__

if __name__ == '__main__':
    data = snap.read(h5ad_input, backed=None)
    group_by = 'class'
    replicate = 'sample'
    n_jobs=30
    print("Started: Find Peaks")
    # Call ATAC-seq peaks using MACS
    snap.tl.macs3(data, groupby=group_by,
                  replicate=replicate,
                  n_jobs=n_jobs)
    print("Finished: Find Peaks")
    print("Started: Merge Peaks")
    merged_peaks = snap.tl.merge_peaks(data.uns['macs3'],
                                                                   chrom_sizes=snap.genome.hg38)
    print("Finished: Merge Peaks")
    print("Started: Make_peak_matrix")
    peak_mat = snap.pp.make_peak_matrix(data, use_rep=merged_peaks['Peaks'])
    print("Finished: Make_peak_matrix")
    print("Started: write_h5ad")
    peak_mat.write_h5ad(output_h5ad, compression="gzip")

Seems to run fine

Started: Find Peaks
100%|██████████| 6/6 [45:43:01<00:00, 27430.32s/it]
... storing 'sample' as categorical
... storing 'class' as categorical
Finished: Find Peaks

Started: Merge Peaks
Finished: Merge Peaks

Started: Make_peak_matrix
Finished: Make_peak_matrix

Started: write_h5ad
Started: write_h5ad

The matrix is empty
AnnData object with n_obs × n_vars = 618362 × 0

Start with the reported entry points snap.tl.macs3, snap.tl.merge_peaks, and snap.pp.make_peak_matrix. Inspect the contents of data.uns['macs3'] and merged_peaks['Peaks'] before matrix creation, then reproduce the run and confirm whether make_peak_matrix produces nonzero variables.