Diamond errors (-6, -7, -11) in bakta 1.12

[fra-cand]

Hello,
I have encountered some problems related to diamond while annotating genomes with bakta latest version (1.12), installed via conda.
In the logs, the diamond error codes are -6, -7 and -11. I saw that -6 and -11 were already reported for previous bakta versions.

File "/home/mibi/miniconda3/envs/bakta/bin/bakta", line 10, in <module>
    sys.exit(main())
             ~~~~^^
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/main.py", line 285, in main
    cdss_psc, cdss_pscc, cdss_not_found = psc.search(cdss_not_found)
                                          ~~~~~~~~~~^^^^^^^^^^^^^^^^
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/psc.py", line 65, in search
    raise Exception(f'diamond error! error code: {proc.returncode}')
Exception: diamond error! error code: -6

File "/home/mibi/miniconda3/envs/bakta/bin/bakta", line 10, in <module>
    sys.exit(main())
             ~~~~^^
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/main.py", line 332, in main
    pseudo_candidates = feat_cds.predict_pseudo_candidates(hypotheticals)
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/features/cds.py", line 621, in predict_pseudo_candidates
    raise Exception(f'diamond error! error code: {proc.returncode}')
Exception: diamond error! error code: -7

Traceback (most recent call last):
  File "/home/mibi/miniconda3/envs/bakta/bin/bakta", line 10, in <module>
    sys.exit(main())
             ~~~~^^
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/main.py", line 285, in main
    cdss_psc, cdss_pscc, cdss_not_found = psc.search(cdss_not_found)
                                          ~~~~~~~~~~^^^^^^^^^^^^^^^^
  File "/home/mibi/miniconda3/envs/bakta/lib/python3.13/site-packages/bakta/psc.py", line 65, in search
    raise Exception(f'diamond error! error code: {proc.returncode}')
Exception: diamond error! error code: -11

I'm annotating a batch of an hundred genomes (downloaded from NCBI) with a for cycle in the terminal, and I'm noticing that these errors seem to occur randomly, they didn't seem to be linked to a particular genome. In fact, if I try to annotate the single genome or if the cycle is relaunched, the annotation may end successfully. I noticed that sometimes the diamond process seems to hang when detecting pseudogenes. It occupies RAM but no CPU, hinderingthe completion of the annotation, so I have to kill it and restart.

I also tried the previous version of bakta (1.11.4) installed via conda with python=3.11 and diamond=2.1.18, but without success. Instead I got a different error:

Traceback (most recent call last):
  File "/home/francesco/miniconda3/envs/bakta_1.11.4/bin/bakta", line 10, in <module>
    sys.exit(main())
             ^^^^^^
  File "/home/francesco/miniconda3/envs/bakta_1.11.4/lib/python3.11/site-packages/bakta/main.py", line 343, in main
    pfam_hits = feat_cds.predict_pfam(hypotheticals)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/francesco/miniconda3/envs/bakta_1.11.4/lib/python3.11/site-packages/bakta/features/cds.py", line 371, in predict_pfam
    aa_identifier = hit.name.decode()
                    ^^^^^^^^^^^^^^^
AttributeError: 'str' object has no attribute 'decode'. Did you mean: 'encode'?

I don't know if this can be somehow fixed since the errors seem to happen randomly, but maybe this report could be useful!
Thanks.

[oschwengers]

Hi, first, thanks for reporting!
So based on this report and other user feedback, I think that this issue is not caused by Bakta but Diamond. Indeed, there are several issue reports with Diamond 2.1.19 & 2.2.20. Therefore, there was another version 2.1.21 released, recently:
https://github.com/bbuchfink/diamond/releases

I hoped that a pin to 2.1.21 would fix most of those reported bugs, and a test on >50 genomes was successful w/o any issue. Hence, I think this is caused by Diamond and needs to be reported upstream.

The 3rd issue that you reported is most certainly due to a wrong PyHMMer version caused by a downgrade of Bakta, This should be fixed with a PyHMMer downgrade to 0.11.4.

[fra-cand]

Thanks a lot! I'll try to downgrade PyHMMer with Bakta 1.11.4 to avoid Diamond's bug.

[oschwengers]

There is also a very new release v2.1.22. You could also give it a try, maybe it helps to fix the issues. Looks very promising due to the release description.

[fra-cand]

I did a small test, bakta 1.11.4 with PyHMMer 0.11.4 seemed to work. No luck again with diamond 2.1.22, the process hanged and didn't proceded. For now I didn't get error codes, but the test was very quick.

[oschwengers]

OK, glad to hear that the Bakta v1.11.4 / PyHMMer v0.11.4 combi works as expected.

Regarding the remaining issue: So if you are sure that Diamond hangs, then this is a clear case of an upstream issue. In this case, I would kindly ask you to file an issue in the Diamond repository, as we unfortunately cannot do anything about it.

[fra-cand]

I'll do some other tests, maybe trying to replicate the issue while using diamond directly and then I'll post on the Diamond repository. Thank a lot!