Cufflinks upper-quantile FPKM generation issue

[TheStudent]

Hi people,

Upper-quartile FPKMs are more robust than 'normal' FPKMs. To generate those I do use the --upper-quartile-norm option in cufflinks (2.02).

Usually the upper-quartile FPKMs should be ~4 higher than the normal FPKMs, but I do get about 1000x higher values.

here an example (consistent over all the genes)

normal FPKMs:
ENSG00000239906 - - ENSG00000239906 RP11-34P13.14 TSS60508 chr1:139789-140339 - - 3.31243 0.349702 6.27515 OK

upper-quartile FPKMs
ENSG00000239906 - - ENSG00000239906 RP11-34P13.14 TSS60508 chr1:139789-140339 - - 103985 10978 196992 OK

Anybody any idea how to solve this? Is this a cufflinks bug?

Thank you very much.

[TheStudent]

upper-quartile of course