Suggestion for improving on annotation transcript start and end positions

[shernadi]

Hello,

I am using Stringtie2 quite frequently to assemble (annotate isoforms) some long-read cDNA data. One thing what I came a cross when I was looking at the results in IGV was that in may cases Stringtie tends to annotate isoforms only with the same start/end positions (usually using the longest option even when there are only a couple of reads supporting that) while the bam file seems to have clear support (with reasonable high coverage) for transcripts with several alternative (shorter) start/end sites. I am wondering if you would have some suggestion for parameters to tweak to try to improve on this?

Thank you!