Dear list, I have a simple doubt concerning how I should deal with the *replicate day* effect using limma. It is clear that I have a *day effect* in my data that it has to be taken into account in the model. I am using 2 different model specifications and I do not know if one of them is correct. a) A first solution is simply to include in the model the *day effect* as an additional fixed effect. I will assume that there is not interaction between *day effect* and *treatment effect*, so my design matrix will be of the form: design=model.matrix(~ - 1 + factor(treatment) + factor(day) ) fit=lmFit(eset,design) b) My question is if I can take into account the *day effect* as random effect using duplicateCorrelation. famrep=c( ) corfit=duplicateCorrelation(eset,ndups=1,block=famrep) design=model.matrix(~ - 1 + factor(treatment)) fit=lmFit(eset,design,block=famrep,cor=corfit$consensus) Would be this second approach a valid one? I will appreciate any comment on this.- Thanks a lot.- Pedro

You could do it either way, and your use of duplicateCorrelation is correct. --Naomi At 07:57 AM 6/28/2006, Pedro L?pez Romero wrote: >Dear list, > >I have a simple doubt concerning how I should deal with the *replicate day* >effect using limma. >It is clear that I have a *day effect* in my data that it has to be taken >into account in the model. I am using 2 different model specifications and I >do not know if one of them is correct. > > >a) A first solution is simply to include in the model the *day effect* as an >additional fixed effect. I will assume that there is not interaction between >*day effect* and *treatment effect*, so my design matrix will be of the >form: > > design=model.matrix(~ - 1 + factor(treatment) + factor(day) ) > > fit=lmFit(eset,design) > > >b) My question is if I can take into account the *day effect* as random >effect using duplicateCorrelation. > > famrep=c( ) > corfit=duplicateCorrelation(eset,ndups=1,block=famrep) > > design=model.matrix(~ - 1 + factor(treatment)) > > fit=lmFit(eset,design,block=famrep,cor=corfit$consensus) > > > Would be this second approach a valid one? > > > > >I will appreciate any comment on this.- >Thanks a lot.- > >Pedro > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor >Search the archives: >http://news.gmane.org/gmane.science.biology.informatics.conductor 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