stringtie (.gtf) vs ballgown (t_data.ctab) FPKM values do not match - why?

Question

[cpak1981]

I'm using stringtie to assemble transcripts. I use stringtie to make a .gtf file, which outputs the following format: ... gene_id "X"; transcript_id "X"; FPKM "0.0111";.... The point is that each entry (transcript/exon) has a FPKM value associated with it (in the Notes field).

I also use stringtie to assemble transcripts into ballgown-compatible files. t_data.ctab also provides FPKM values for each transcript.

My issue is that these FPKM values (for a specific transcript) do not match...has anyone else compared these values? Any idea why there's a difference? And which value I should trust?

[lakhujanivijay]

Please provide examples of differences