I'm trying to find novel isoforms from RNA-Seq. I have six samples, 3 control and 3 treatment. I made a merged gtf file after running StringTie with gencode annotation. Then the merged annotation was used to estimate FPKM in all samples. Then novel transcripts expression was tabulated as shown below. The left three columns are control and the right three are treatment. Why is the fpkm absolute zero for in some cases, when it is very high for the same transcript in other samples.