[BioC] gls.series error ?

Gordon K Smyth smyth at wehi.EDU.AU
Sat Mar 6 07:02:57 MET 2004


> Hi,
>
> I use limma to analyse my dye swap. I have 3 replicates with a spacing
> of 1152.
> When I use the gls.series function I obtain a very small correlation.
> fit <- gls.series(data,design=flip,ndups=3,spacing=1152)
> fit$cor : 0.1937247
> But if I use the dupcor function
> cor <- dupcor.series(data,design=flip,ndups=3,spacing=1152)
> cor$cor = 0.6515627
> if I "miss" the spacing
> cor <- dupcor.series(data,design=flip,ndups=3,spacing=1)
> cor$cor = 0.1937247
>
> The gls.series function don't use the spacing argument to calculate
> correlation ?

Woops, it's a bug.  Thanks for pointing this out.  I'll fix it.

Gordon

> Thanks
> Valérie



More information about the Bioconductor mailing list