I was working with a given GFF3 file, and I observed that some transcripts have orientation opposite to their transcripts. Here is a snippet:
chr1 . gene 2189548 2194772 . + . ID=c5eaf8c1-2bb2-566f-e154-6a9be797c11f;Name=FAAP20;gene_biotype=protein_coding;gene_id=34a96673-bfab-ebbf-b2b7-f9cb73163a4a;gene_name=FAAP20
chr1 . transcript 2189548 2194772 . - . ID=f47feeac-74ce-1543-53b6-afd9bbb8ab19;Name=rna-NM_182533.4;gene_biotype=protein_coding;gene_id=34a96673-bfab-ebbf-b2b7-f9cb73163a4a;gene_name=FAAP20;Parent=c5eaf8c1-2bb2-566f-e154-6a9be797c11f;transcript_biotype=protein_coding;transcript_name=rna-NM_182533.4;protein_id=NP_872339.3
How do I interpret this? I checked UCSC Browser and in each case it seems that the browser agrees with the transcript in each case, which seems like it should be canonical.
Moreover, I observed in this file that gene orientation is always +. Whereas a highly curated GFF3 of the human genome represents both directions:
# my file
$ awk '$3 == "gene" {print}' filtered_potential_genes.gff | cut -f7 | sort | uniq -c
1600 +
# highly curated annotations
$ awk '$3 == "gene" {print}' /MANE.GRCh38.v0.95.select_refseq_genomic.gff | cut -f7 | sort| uniq -c
9393 +
9191 -
According to the person who provided me the file, the existence of bidirectional transcription was indeed the motivation for this. However, it's obviously confusing to have it still be a definitive + rather than . or something to signify unstrandedness.
It seems likely that this is due to an artifact in my file. Are there other interpretations? Is the standard simply that gene orientation is meaningless and transcript orientation is important?
