Entering edit mode
Dear Jakob,
This doesn't necessarily mean that there's
anything wrong with your linux installation, and
the limma versions for Windows and linux are
identical. There are differences in floating
point arithmetic between different hardward
architectures, and you've run into a flag in one
of the Lapack routines which apparently doesn't get tripped under
Windows.
You could try to identify which probe, or maybe
probes, is causing the error in linux and remove
it from the fit. Or else you could simply use the
duplicateCorrelation consensus correlation you got from Windows.
Best wishes
Gordon
>[BioC] Limma - error when using duplicateCorrelation - Windows ver.
Linux
>Jakob Hedegaard Jakob.Hedegaard at agrsci.dk
>Tue Jan 16 15:19:38 CET 2007
>
>Hi List
>
>When using duplicateCorrelation in Limma running
>under Linux I am receiving the following error:
>
> > cor <- duplicateCorrelation(MA, design=design, ndups=2,
weights=MA$weights)
>Error in La.svd(QtZ, nu = mq, nv = 0) : error
>code 1 from Lapack routine 'dgesdd'
>
>Interestingly, there are no problems when
>analysing the same dataset in Limma running
>under Windows indicating a problem with the Linux
version/installation....
>
>- any suggestions would be appreciated!
>
>Linux:
> > sessionInfo()
>Version 2.3.1 (2006-06-01)
>x86_64-unknown-linux-gnu
>
>attached base packages:
>[1] "methods" "stats" "graphics" "grDevices" "utils"
"datasets"
>[7] "base"
>
>other attached packages:
> statmod limma
> "1.2.4" "2.7.10"
> >
>
>Windows:
> > sessionInfo()
>Version 2.3.1 (2006-06-01)
>i386-pc-mingw32
>
>attached base packages:
>[1] "methods" "stats" "graphics" "grDevices" "utils"
"datasets"
>[7] "base"
>
>other attached packages:
> statmod limma
> "1.2.4" "2.7.10"
> >
>
>
>
>Best regards
>
>Jakob Hedegaard
>Project scientist
>
>
>UNIVERSITY OF AARHUS
>Faculty of Agricultural Sciences
>Research Centre Foulum
>Dept. of Genetics and Biotechnology
>Blichers All? 20, P.O. BOX 50
>DK-8830 Tjele