This is likely due to the upgrade from Cufflinks 1.3.x to Cufflinks 2.0.x; Cufflinks 2.0 introduced a new algorithm for Cuffdiff in particular. You can read about these changes on the website:
http://cufflinks.cbcb.umd.edu/ (and there's a manuscript describing the changes as well).

You might consider writer to to the tool authors directly for more details: tophat.cufflinks@gmail.com Of course, please consider sharing anything you learn with members of this list as well.

Best,
J.



On Mar 13, 2013, at 12:06 PM, Mohammad Heydarian wrote:

We are having the exact same issue, on the main server and our (recent) cloud instances.

Were some of the hidden Cuffdiff parameters modified since fall 2012?

Cheers,
Mo Heydarian

On Mar 13, 2013 11:02 AM, "Jenna Smith" <jes227@case.edu> wrote:
Hi,

I'll preface my concern by saying that I'm a novice to Cufflinks.  Back in September, I performed a Cuffdiff analysis comparing a wild-type and mutant condition.  The analysis returned ~800 transcripts differentially regulated between the two with statistical significance.  Recently, I've rerun the Cuffdiff analysis - using exactly the same files stored in Galaxy for all inputs, and with all the same parameters - and only get a few dozen statistically significant hits.  However, all of the data besides the p and q values are essentially identical between these two runs, so I am really unclear as to what is causing the difference.  Here is just one clear example:

From run 1:
YFR026C
FPKM 1 = 17.2434
FPKM 2 = 196.735
log2(fold change) = 3.51214
p = 1.64E-8
q = 7.33E-6
significant = yes

From run 2:
YFR026C
FPKM 1 = 14.4489
FPKM 2 = 144.939
log2(fold change) = 3.32641
p = 0.000170034
q = 0.0719964
significant = no

The second Cuffdiff analysis shows there is still a ~10-fold difference between conditions, but this is not statistically significant.  Has the version of Cuffdiff on Galaxy been updated such that some parameters have changed, that could explain this difference?  Or, is there some setting I am missing that would cause very large changes to fail statistical significance testing?  Any help or input would be appreciated, I am really at a loss for why executing what should be exactly the same task is giving vastly different results.  I could just be overlooking something very fundamental that is obvious to someone with more experience with this program.  Thanks.

-Jenna Smith


___________________________________________________________
The Galaxy User list should be used for the discussion of
Galaxy analysis and other features on the public server
at usegalaxy.org.  Please keep all replies on the list by
using "reply all" in your mail client.  For discussion of
local Galaxy instances and the Galaxy source code, please
use the Galaxy Development list:

  http://lists.bx.psu.edu/listinfo/galaxy-dev

To manage your subscriptions to this and other Galaxy lists,
please use the interface at:

  http://lists.bx.psu.edu/
___________________________________________________________
The Galaxy User list should be used for the discussion of
Galaxy analysis and other features on the public server
at usegalaxy.org.  Please keep all replies on the list by
using "reply all" in your mail client.  For discussion of
local Galaxy instances and the Galaxy source code, please
use the Galaxy Development list:

 http://lists.bx.psu.edu/listinfo/galaxy-dev

To manage your subscriptions to this and other Galaxy lists,
please use the interface at:

 http://lists.bx.psu.edu/