Internal KeyError when processing externally generated gene calls about gunc

[jdwinkler-lanzatech]

Think I figured it out. It looks like the basename extracted from the diamond output file will include the original extension, while the gene counts are stored in the filename without the extension. Running GUNC using gene calls without an extension (e.g. proteins) works fine.

[fullama]

Yeah i just saw that too.. i can only apologise for that. I will release a new version with a fix as soon as i can, at least for now you could run without an extension. Thanks for pointing it out though.. i dont know how that got through testing..

[jdwinkler-lanzatech]

No worries, happens to all of us.

[fullama]

I just released v1.0.3 which should fix this issue.. it is taking a while to appear on conda but once its there this issue should be solved!

if you notice any other issues do let me know.. thanks again.

[cpauvert]

Hi @fullama,
I am reviving this issue because I could reproduce the same error as @jdwinkler-lanzatech in GUNC v.1.0.5.
For externally generated gene calls to work, one must still indicate the (.faa) FASTA file without the extension. This can be problematic for workflow management system based on the filenames to infer dependencies (e.g. Snakemake).

As far as I have understood, there are two possibilities of fix:
A. keeping the prefix (.faa) in the genes_counts json file (and modify here)
B. Adding a specific condition (if gene_calls) to the extraction of the basename here

The A solution could be complemented by a last read of the json, trimming of the suffixes and rewrite, for compatibility purpose.
Let me know your thoughts and keep up the good work!
Best,

[fullama]

hi, im not sure i understand whats happening here.. did you run with --file_suffix .faa ?

[cpauvert]

My bad, it worked with --file_suffix .faa
Thanks @fullama for the quick reply.