@Thomas Danhorn Thank you for your response. You are not misunderstanding anything. Before I saw this I would believe that it is impossible. I am quite surprised I could find more information about this, given this is a rice genome, which is a model organism thanks to its short genome and economic importance. But NCBI has tooling to show this with

complement()

, so I would assume that it is not complete nonsense. And I also assume that given its not just some random seq, but a genome marked as a reference by NCBI, that NCBI checked that it is not boohafle. And I believe the annotation was created by NCBI as well. I checked some older genomes on NCBI and they typically don't even have mitochondrion sequenced or if they do, not annotated. Do you know if there is perhaps a better venue where I could ask about this?

I don't think that gene is encoded in the mitochondria (although the gene product might be located there), otherwise it would not be "spliced" (but give how weird this is, I should be careful with conclusions). Venues for asking general bioinformatics questions are BioStars and SeqAnswers, but I doubt anyone there will have an authoritative answer for something so outlandish. If you want to know what NCBI was thinking when they created this, you best bet is to ask them. I once had a (admittedly easier) question for them and got a very competent and helpful answer. I believe I used a web form to submit my RefSeq-related request, and the reply came from nlm-support@nlm.nih.gov. I have my doubts that asking will give you any great new options (unless there is an alternate reference genome), and I have already given you my suggestions for work-arounds. 🤞