As with 3'end, StringTie is likely not defining these very accurately (is fairly clear from manual browsing but I shouldn't jump to conclusions...)
As this is not a problem I'm trying to solve (defining novel 5'ends/TSS), it may be worth avoiding adding extra noise/variation/uncertainty by replacing the assembled first exons of novel events with the matching reference first exon.
As with 3'end, StringTie is likely not defining these very accurately (is fairly clear from manual browsing but I shouldn't jump to conclusions...)
As this is not a problem I'm trying to solve (defining novel 5'ends/TSS), it may be worth avoiding adding extra noise/variation/uncertainty by replacing the assembled first exons of novel events with the matching reference first exon.
Something to consider...