I am running macs3 hmmratac on the ATAC-seq bam files with the command below:

 macs3 hmmratac --cutoff-analysis-only -i sample.bam -f BAMPE 

and getting the following the error:

Random seed selected as: 10151
Use --hmm-type to select a Gaussian ('gaussian') or Poisson ('poisson') model for the hidden markov model in HMMRATAC. Default: 'gaussian'.

INFO  @ 08 Jun 2025 21:31:10: [107 MB] #1 Read fragments from BAMPE file...
INFO  @ 08 Jun 2025 21:31:15: [144 MB]  1000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:18: [153 MB]  2000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:23: [154 MB]  3000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:27: [167 MB]  4000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:31: [170 MB]  5000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:36: [178 MB]  6000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:40: [186 MB]  7000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:45: [193 MB]  8000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:49: [199 MB]  9000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:53: [215 MB]  10000000 fragments parsed
INFO  @ 08 Jun 2025 21:31:57: [225 MB]  11000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:00: [235 MB]  12000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:04: [244 MB]  13000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:08: [246 MB]  14000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:12: [256 MB]  15000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:15: [273 MB]  16000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:19: [282 MB]  17000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:23: [293 MB]  18000000 fragments parsed
INFO  @ 08 Jun 2025 21:32:26: [294 MB] 18913216 fragments have been read.
INFO  @ 08 Jun 2025 21:32:34: [294 MB] #  Read 18913216 fragments.
INFO  @ 08 Jun 2025 21:32:34: [294 MB] #2 Use EM algorithm to estimate means and stddevs of fragment lengths
INFO  @ 08 Jun 2025 21:32:34: [294 MB] #  for mono-, di-, and tri-nucleosomal signals...
INFO  @ 08 Jun 2025 21:32:34: [294 MB] # A random seed 10151 has been used in the sampling function
INFO  @ 08 Jun 2025 21:32:35: [328 MB] # Downsampled 1594057 fragments will be used for EM training...
 ValueError:  Adjust --means and --stddev options and re-run command

After alignment, I have only considered the reads mapped to Chr1-22, and X and Y, followed by marking duplicates using Picard, using samtools with -f 2 -F 1548 -q 30 for filtering and finally usingbedtools intersect -nonamecheck -v -abam to remove the blacklisted regions. What could be the reasons for the error at hmmratac step? I tried running macs3 hmmratac on the bam file obtained just after the alignment without involving any downstream filtering steps, and it runs absolutely fine without any error.