Dear list, I tried to create a new exprSet from my filtered Affy genelist using R 2.3.0 with affy "1.10.0" simpleaffy "2.6.0" genefilter "1.10.0" and I got the folowing error: good.eset = new("exprSet", exprs = good) Error in validObject(.Object) : invalid class "exprSet" object: number of exprs columns different from number of pData rows But dim(good) [1] 7420 20 dim(pData(data)) [1] 20 1 I moved then to R version 2.2.0, 2005-10-06 affy "1.8.1" simpleaffy "2.4.2" genefilter "1.8.0" and the very same commands gave me no error. >good.eset = new("exprSet", exprs = good) >good.eset Expression Set (exprSet) with 7420 genes 20 samples phenoData object with 0 variables and 0 cases varLabels Have I missed some critical changes in the new versions of the packages?

The change is in the Biobase package, which affy et al rely on. Biobase now checks to make sure the objects it creates are valid, in this case that the number of exprs columns (i.e., samples) equals the number of pData rows. When you say > good.eset = new("exprSet", exprs = good) exprSet sees that it's supposed to create an instance with 20 samples, but how can it create the corresponding phenoData? From this: > dim(pData(data)) it looks like you have a phenoData object that you want to associate with your exprs data. So... good.eset = new("exprSet", exprs=good, phenoData=data) (perhaps data is an exprSet, and you need to provide phenoData(data) as the value of phenoData) should do the trick. Notice that in the old version >good.eset = new("exprSet", exprs = good) >good.eset Expression Set (exprSet) with 7420 genes 20 samples phenoData object with 0 variables and 0 cases varLabels you've generated an invalid exprSet -- no description of the phenotypes associated with your samples! Hope that helps Martin kfbargad at ehu.es writes: > Dear list, > > I tried to create a new exprSet from my filtered Affy genelist using R > 2.3.0 with > affy "1.10.0" > simpleaffy "2.6.0" > genefilter "1.10.0" > > and I got the folowing error: > > good.eset = new("exprSet", exprs = good) > Error in validObject(.Object) : invalid class "exprSet" object: number > of exprs columns different from number of pData rows > > But > dim(good) > [1] 7420 20 > dim(pData(data)) > [1] 20 1 > > I moved then to R version 2.2.0, 2005-10-06 > affy "1.8.1" > simpleaffy "2.4.2" > genefilter "1.8.0" > > and the very same commands gave me no error. >>good.eset = new("exprSet", exprs = good) >>good.eset > Expression Set (exprSet) with > 7420 genes > 20 samples > phenoData object with 0 variables and 0 cases > varLabels > > > Have I missed some critical changes in the new versions of the > packages? > > _______________________________________________ > 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

Peter <bioconductor-mailinglist at="" maubp.freeserve.co.uk=""> writes: > If you don't have any real phenotype data, then this will generate a > dummy phenotype object with the sample number: > > pheno <- read.phenoData(NULL, colnames(data), FALSE) > good.eset <- new("exprSet", exprs=good, phenoData=pheno) > > The resulting phenotype data is just a "sample number" (i.e. column > number in the data matrix/array). One of the main ideas of the exprSet class is to organize expression values _and_ sample data (phenotype info). In other words, I don't think the common case is one in which there is no phenotype data other than an ID. As you have found, the case of ID only can be handled by the current code. If there are others who would find such a default useful, please speak up, otherwise I would suggest a smaller helper function to create new exprSet instances for your particular use case. Best, + seth