StringTie gives incorrect 5 end.

[Pentayouth]

Dear gpertea and all Stringtie maintainer,

I performed single sample assembly with StringTie however the reconstructed 5 end is incorrect even the read coverage support in bigwig is obvious.

The parameters I used are --rf -p 1 -j 2 -c 2 -s 5 -f 0.05 (without -G because I perfer de novo assembly)

Here is the example of gene KRT18:

The grey track is gencode v40 reference gtf. The red arrow denotes the transcript which I think should be reconstructed by StringTie

However, in all single sample assemblies, StringTie gave a shorter, more downstream 5 end.

I couldn't figure out why could it happen. Is there anything to do with my parameter?

Best regards,
Wang