All, In section 8.6.2 of the limma user's guide, an example is given using splines for time-course data. It looks like in this example, the data points are independent, meaning different subjects are observed at different timepoints. If the same subject is observed over time, is it correct to use duplicateCorrelation function along with the spline model. Is this the correct way to handle profiles of individuals in limma? What other Bioconductor approaches have people used? I can't tell if EDGE (Leek and colleagues) is updated/maintained. Thanks, Juliet [[alternative HTML version deleted]]

For repeated measures of the same individual you will want to include an identified as a factor in the model. Eg. where there are 2 patients "p1" and "p2". ids <- factor(c("p1", "p1", "p1", "p2", "p2", "p2")) design <- model.matrix(~yourExistingModel+ids) Paul. On Tue, Apr 16, 2013 at 2:10 PM, Juliet Hannah <juliet.hannah at="" gmail.com=""> wrote: > All, > > In section 8.6.2 of the limma user's guide, an example is given using > splines for time-course data. It looks like in this example, the data > points are independent, meaning different subjects are observed at > different timepoints. > > If the same subject is observed over time, is it correct to use > duplicateCorrelation function along with the spline model. Is this the > correct way to handle profiles of individuals in limma? > > What other Bioconductor approaches have people used? I can't tell if EDGE > (Leek and colleagues) is updated/maintained. > > Thanks, > > Juliet > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- Dr. Paul Geeleher, PhD (Bioinformatics) Section of Hematology-Oncology Department of Medicine The University of Chicago 900 E. 57th St., KCBD, Room 7144 Chicago, IL 60637 -- www.bioinformaticstutorials.com

Dear Juliet, Yes, using duplicateCorrelation to estimate the within-subject correlation is a valid way to go. The alternative is to set subject as an explanatory factor as Paul Geeleher suggested in his reply. The first approach is statistically more powerful, the second makes fewer assumptions. If you have lots of subjects, first approach might be good. If you have only two subjects or some subjects are distinctly different from others (outliers), use the second approach. Best wishes Gordon > Date: Tue, 16 Apr 2013 15:10:04 -0400 > From: Juliet Hannah <juliet.hannah at="" gmail.com=""> > To: Bioconductor mailing list <bioconductor at="" r-project.org=""> > Subject: [BioC] spline example from limma user's guide > > All, > > In section 8.6.2 of the limma user's guide, an example is given using > splines for time-course data. It looks like in this example, the data > points are independent, meaning different subjects are observed at > different timepoints. > > If the same subject is observed over time, is it correct to use > duplicateCorrelation function along with the spline model. Is this the > correct way to handle profiles of individuals in limma? > > What other Bioconductor approaches have people used? I can't tell if EDGE > (Leek and colleagues) is updated/maintained. > > Thanks, > > Juliet > ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}