cuffmerge FPKM values

Question

[zauhar]

I see a number of questions about cuffmerge, asking for example why FPKMs are missing.

In my case cuffmerge IS producing FPKMs, and they are a small fraction of the values reported in the transcripts files for the individual samples. Is everything going to be re-done from the original samples when running cuffdiff? If so, why does cuffmerge even include an FPKM?

Thanks!