ZeroDivisionError: float division by zero

[f12345zxcvbnm2]

Summary:
ZeroDivisionError: float division by zero

Description:
I am trying to use average_nucleotide_identity.py script for the dataset of bacteria genomes,
my command was:
average_nucleotide_identity.py -i ./0.6_test/ -o ./7_ANI -m ANIm -g -v

Traceback (most recent call last):
  File "/data/h01003/miniconda3/envs/pyani/bin/average_nucleotide_identity.py", line 10, in <module>
    sys.exit(run_main())
  File "/data/h01003/miniconda3/envs/pyani/lib/python3.8/site-packages/pyani/scripts/average_nucleotide_identity.py", line 998, in run_main
    results = method_function(args, infiles, org_lengths)
  File "/data/h01003/miniconda3/envs/pyani/lib/python3.8/site-packages/pyani/scripts/average_nucleotide_identity.py", line 559, in calculate_anim
    results = anim.process_deltadir(deltadir, org_lengths, logger=logger)
  File "/data/h01003/miniconda3/envs/pyani/lib/python3.8/site-packages/pyani/anim.py", line 480, in process_deltadir
    ) = parse_delta(deltafile)
  File "/data/h01003/miniconda3/envs/pyani/lib/python3.8/site-packages/pyani/anim.py", line 396, in parse_delta
    avrg_ID = sum(weighted_identical_bases) / sum(aligned_bases)
ZeroDivisionError: division by zero

pyani Version:
pyani 0.3.0

Python Version:
Python 3.8.19

Operating System:
Ubuntu 20.04.6 LTS

here are my files:
2_bacteria.zip

[peterjc]

Your two bacteria are so different they have no meaningful alignment in nucmer, so the ANIm value ought to be reported as null or NA. Or perhaps zero.

Practically speaking I guess you have probably tried this on a larger set and narrowed it down to a small test case? If so, you've already identified an outlier best left out of the analysis.

[f12345zxcvbnm2]

Greetings, thank you very much for your response. These two genomes were specifically selected from 300 genomes because I reviewed the nucmer_output folder. Now, the situation is a bit tricky. I need to identify and filter out the genomes that cannot be calculated from the entire dataset. I was wondering if pyani has this filtering function? If not, it would be incredibly helpful if such a filtering feature could be included in future updates.

[peterjc]

The team is aware of this issue and thinking about how to best handle it as part of a larger engineering effort rewriting pyANI: https://github.com/pyani-plus/pyani-plus

Note pyANI-plus is still unfinished and has not had a public release, but for this specific example the failed comparisons are handled gracefully with a null recorded in the database. This does currently limit the downstream analysis options - see pyani-plus/pyani-plus#272 in particular. If you fancy trying this out, some early end-user feedback would be welcome... but there is no end-user facing documentation as yet beyond the command line help.