[BioC] lmFit and block argument for technical replicates

[Naomi Altman]

I've been doing it the 2nd way.

--Naomi

At 02:53 PM 11/12/2005, Carolyn Fitzsimmons wrote:


>Hello,
>
>I have become confused about the use of the block argument in the
>lmFit function of limma when dealing with technical replicates.
>
>I have a common reference design with 38 slides (19 cDNAs x 2 replicates).
>
>Previously I used lmFit in this manner:
>####################################
>  > cor <- duplicateCorrelation(Mmatrix_ny, design=design,
>+
>block=c(1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,19,19)
>+ )
>   > cor$cor
>[1] 0.06795489
>  > fit <- lmFit(Mmatrix_ny, design=design, correlation=cor$cor)
>####################################
>But after re-reading the limma user's guide again I believe the lmFit
>line should read:
>####################################
>  > fit <- lmFit(Mmatrix_ny, design=design,
>block=c(1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,19,19), 
>
>correlation=cor$cor)
>####################################
>
>But I am still uncertain.  Which is correct?
>
>Thanks for your time,  Carolyn
>
>
>         [[alternative HTML version deleted]]
>
>_______________________________________________
>Bioconductor mailing list
>Bioconductor at stat.math.ethz.ch
>https://stat.ethz.ch/mailman/listinfo/bioconductor

Naomi S. Altman                                814-865-3791 (voice)
Associate Professor
Dept. of Statistics                              814-863-7114 (fax)
Penn State University                         814-865-1348 (Statistics)
University Park, PA 16802-2111