ValueError: max() arg is an empty sequence

[sc-wenxinqiang]

Hi GFF3toolkit developers. I'd like to seek for your help on this issue. QC part went well, but running the fix step raises a ValueError exception. For reference, I'm attaching the gff file and the error.txt here.
error.zip
evm.zip
fix.zip
Thank you.

[mpoelchau]

Thanks for reaching out @sc-wenxinqiang ! Sorry that the error is so uninformative. The problem is that gff3-fix can't fix the Emr0003 errors (duplicate IDs), and those need to be fixed for it to handle the Ema0001 errors (Parent feature start and end coordinates exceed those of child features).

I looked at the gff file. How was this generated? It looks like most lines are duplicated and are out of order. I recommend that you re-generate it if you can - I'd be wary of having another program fix it if the gff file itself is broken to this degree.

[jaspwn]

Hi,

I am also getting the same error message but I am not getting an Ema0001 errors. Is this a list somewhere of which errors cannot be fixed by gff.fix? I have attached my error file here.

QC.out.gz

[mpoelchau]

@jaspwn you can find the documentation on gff3-fix here, including what errors can't be fixed: https://github.com/NAL-i5K/GFF3toolkit/blob/master/docs/gff3_fix.py-documentation.rst

Looks like you have Ema0002 (Protein sequence contains internal stop codons), Ema0004 (Incomplete gene feature that should contain at least one mRNA, exon, and CDS), and Ema0008 (Warning for distinct isoforms that do not share any regions) errors, which can't be fixed by gff3-fix.

[jaspwn]

Ah thanks, I should have looked a little harder (-:

Can I also ask, do you know of any tool that allows a semi-automated way of renaming individual annotations?

For example if I had a list of apollo-like gene ID numbers, I would want to manually rename the gene along with the subsequent children associated with that gene (CDS, exons, UTRs etc.)

[mpoelchau]

Our ancillary script gff3tool/lib/gff3_ID_generator.py can do this. It has some arguments related to gff3-merge but you can ignore those. So if you'd want to rename IDs with a prefix tmp and 7 digits, you'd do python gff3tool/lib/gff3_ID_generator.py -g INPUT.gff -og OUTPUT.gff -idpre tmp -diglen 7 -r RENAME-report.txt. When I run this I activate my python environment where I've installed the gff3toolkit first.

You could also look into agat, I think it's more actively maintained than our toolkit ;). https://agat.readthedocs.io/en/latest/