gwct
commented
1 year ago So I can replicate this exactly, though one thing I will point out is that degenotate looks for either "transcript" or "mRNA" and "CDS" (not "exon"). But this problem is happening when the transcripts are read so we just haven't gotten to any issues with your switch from "CDS" to "exon".
Basically, when we just swap the "gene" string for the "transcript" or "mRNA" string in column 3 of the gff, this leads to a host of problems when parsing the parent ID of the transcript which doesn't exist since it was annotated as a "gene" feature -- normally the parent of a transcript is a gene. degenotate uses this mainly to determine longest transcript per gene, I think, but this check for the parent ID is kind of baked in to the gff parsing code so it will take a bit of effort to get it out such that this can be disabled.
The real solution is for GFFs to be formatted in a more standard way. I have no idea why this one doesn't have transcript features. Of course this isn't a helpful solution for your problem at hand. Unless @tsackton has any other suggestions, this will take a bit of re-factoring this code to solve. I could see us either having an option to disable the parent look-up for transcripts if the user doesn't care about longest transcripts or, probably less ideally, another read-through of the gff to determine if it has "transcript" features at all, and if not to fall back to "gene" features as transcripts with the parent look-up disabled.
tsackton
wutian1217
Hello,
I got interesting errors when testing degenotate v1.1.2 on the assembly and annotation for Eleusine coracana, which you can download here.
The command:
The output:
I think the problem is that this GFF variant has only "gene" and "CDS" features, whereas I think degenotate looks for "mRNA" and "exon" features, correct? I tried to fix this error by changing genes and CDS to mRNA and exons, respectively
sed 's/\tgene/\tmRNA/' data/annotations/eleusine_coracana.gff3 | sed 's/\tCDS/\texon/' > tmp.gff3But I then got a different error about ID parsing
The new command:
python degenotate.py --overwrite -d " " -a tmp.gff3 -g data/assemblies/eleusine_coracana.fa -o data/degenotateOutput/eleusine_coracanaThe new output:
I'm not quite sure what this error refers to because degenotate should be able to properly construct transcripts by looking at the matching ID and Parent tags of the mRNA and exon features, respectively. But maybe I'm missing something about how degenotate behaves.
Anyway, I wanted to post here because I didn't see either of these errors come up in the issue section yet.
Thanks in advance for your help! And let me know if you need more info!