[BioC] simple ANOVA

Roger Vallejo rvallejo at psu.edu
Wed Nov 26 20:12:21 MET 2003

I have a very simple design: it has 2 tissues (saline vs ecoli
injected), 2 time-point evaluation (24 hr and 48 hr post injection) with
2 observations per experimental unit (2 reps). We are using Affymetrix
genechip arrays (mouse expression set 430A), so it makes a total of 8
arrays. It is a very simple ANOVA that I am trying to analyze using the
R-package. I am using a UNIX SunOS 5.8, Solaris 8.0. I have followed the
LIMMA vignette, and I got this error message: "No residual degrees of
freedom in linear model fits". 

These are the commands that I used:


Limma 11/24/03 (did not work why?):

> The usual: change directory in R-window

> library(affy)

> Data <- ReadAffy()

> Data

> eset <- expresso(Data,bgcorrect.method="rma",
normalize.method="quantiles", pmcorrect.method="pmonly",

> exprs(eset)            ## runs for awhile, 5 minutes for 22K

> design <- model.matrix(~ -1+factor(c(1,2,1,2,3,4,3,4)))        ## I
have array1-8 (4 RNAs and 2 reps)

> colnames(design) <- c("group1", "group2", "group3", "group4")  ## I
assume the groups refer to RNAs

> library(limma)

> fit <- lmFit(eset, design)                                 ## works

> contrast.matrix <- makeContrasts(group2-group1, group4-group3,

> fit2 <- contrasts.fit(fit, contrast.matrix) 
> fit2 <- eBayes(fit2)                                            ##
here I got error messages "no residual df in model fits"



I would appreciate having some help on solving this problem.

Thanks a lot.




Roger L. Vallejo, Ph.D.

Assist. Professor of Genomics/Bioinformatics

The Pennsylvania State University

Department of Dairy & Animal Science

Genomics & Bioinformatics Laboratory

305 Henning Building

University Park, PA 16802

Phone:        (814) 865-1846 

Fax:            (814) 863-6042

Email:         rvallejo at psu.edu <mailto:rvallejo at psu.edu> 

Website:     http://genomics.cas.psu.edu/ <http://genomics.cas.psu.edu/>


	[[alternative HTML version deleted]]

More information about the Bioconductor mailing list