Cufflinks FPKM output not consistent with different thread numbers

[Rachel Hillmer]

Hi,

I find that the FPKM values cufflinks prints out when I run it with multiple threads (p -5) are about 4 orders of magnitude higher than when I run cufflinks in a single-processor mode (p -1, or unspecified since 1 is the default).

Is anyone else having this problem?

This happens regardless of which cufflinks version I'm using (1.3.0 or 2.0.2).

Cheers,
~Rachel

[Rachel Hillmer]

The above is an incorrect report. I happened to also be changing whether or not quartile normalization was being used concurrently with additional threads. Further investigation reveals that this huge change in FPKMs (~4 orders of magnitude) is directly linked to quartile normalization. That's still confusing, but I'm starting a new thread for that, with a more appropriate title.