Failure downloading certain busco lineages

[edwhisnant]

I am running into an issue with busco recognizing and downloading certain lineages during the training module. The two I have come across are 'aspergillaceae' and 'thelebolales'. When I run A. nidulans or A. niger, there is no issue -- as the lineage is for 'aspergillus'.
Here is an example of the error:

Genome: Pseudogymnoascus_verrucosus (GCA_001662655.1) (Thelobolales; Leotiomycetes)):

Choosing best busco species based on taxonomy: thelebolales�[0m
Traceback (most recent call last):
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/bin/funannotate2", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/lib/python3.12/site-packages/funannotate2/__main__.py", line 28, in main
    train(args)
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/lib/python3.12/site-packages/funannotate2/train.py", line 148, in train
    busco_url = download_urls["busco"][busco_species][0]
                ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
KeyError: 'thelebolales'

Monascus_ruber (GCA_002976275.1) (Aspergillaceae):

Choosing best busco species based on taxonomy: aspergillaceae�[0m
Traceback (most recent call last):
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/bin/funannotate2", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/lib/python3.12/site-packages/funannotate2/__main__.py", line 28, in main
    train(args)
  File "/hpc/group/bio1/ewhisnant/miniconda3/envs/funannotate2/lib/python3.12/site-packages/funannotate2/train.py", line 148, in train
    busco_url = download_urls["busco"][busco_species][0]
                ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
KeyError: 'aspergillaceae'

[nextgenusfs]

Thanks for reporting. Need to check upstream if the urls/links have moved.

[nextgenusfs]

Okay, so there is no thelebolales or aspergillaceae in BUSCO, investigating why algorithm selected those.

[edwhisnant]

Thanks for checking... I was able to make a work around by explicitly using the best taxonomic ID in the --species argument, which bypassed the problem temporarily. But weird that they were selected, even though those databases don't exist.

[nextgenusfs]

Okay, was a bug in the codebase, this is now working I think:

>>> from funannotate2.utilities import *
>>> from funannotate2.config import *
>>> taxonomy = lookup_taxonomy('Monascus ruber')
>>> choose_best_augustus_species(taxonomy)
'aspergillus_fumigatus'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_oryzae'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_oryzae'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_fumigatus'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_oryzae'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_fumigatus'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_oryzae'
>>> choose_best_augustus_species(taxonomy)
'aspergillus_nidulans'
>>> choose_best_busco_species(taxonomy)
'eurotiales'
>>> taxonomy = lookup_taxonomy('Pseudogymnoascus verrucosus')
>>> choose_best_augustus_species(taxonomy)
'botrytis_cinerea'
>>> choose_best_augustus_species(taxonomy)
'botrytis_cinerea'
>>> choose_best_augustus_species(taxonomy)
'botrytis_cinerea'
>>> choose_best_busco_species(taxonomy)
'leotiomycetes'

the Augustus species is random when there is multiple taxonomic levels that are closest. Shouldn't have a big impact, it's just setting the priors for training.

PS, nice to see somebody working with Pseudogymnoascus -- I might know a guy who submitted that genome ;)