Hello,
I am using stringtie to predict transcripts from long reads. Overall it works well, but some of the predicted transcripts appear to be two genes fused together, due to overlapping UTRs by a small number of reads with run-on UTRs. This does not account for very many reads overall, though still has the result that two adjacent genes get fused into a single transcript or gene.
This occasionally even happens if two reads are pointed 3' to 3'.
Is there some way to be stricter about how many long reads are needed before joining transcripts into genes, and is there a way to require that they are on the same strand? What would be the best way of doing this?
Additionally, it may be helpful in the manual if -L was clarified better, perhaps by saying -L use long reads settings, equivalent to -g 0 -s 1.5 (default:false) or something similar.
Hello, I am using stringtie to predict transcripts from long reads. Overall it works well, but some of the predicted transcripts appear to be two genes fused together, due to overlapping UTRs by a small number of reads with run-on UTRs. This does not account for very many reads overall, though still has the result that two adjacent genes get fused into a single transcript or gene.
This occasionally even happens if two reads are pointed 3' to 3'.
Is there some way to be stricter about how many long reads are needed before joining transcripts into genes, and is there a way to require that they are on the same strand? What would be the best way of doing this?
Additionally, it may be helpful in the manual if -L was clarified better, perhaps by saying
-L use long reads settings, equivalent to -g 0 -s 1.5 (default:false)or something similar.Thank you wrf