Hi: I thought I would like to share a situation that I have been facing. It is not a bug but something that happens in a specific condition. When you use the "medianpolish" method in "expresso" and also use the "subtractmm" option in the "pmcorrect" option you will find that the method will blow up. The reason for this is sometimes the PM-MM will yield "negative" values and this will cause the "NaN" values. I guess is it good to have the ability to choose the transformation function to something other than log2 for negative values. Any suggestions/thoughts on this. Thanks Shyam

the affy package provides the infrastucture so that you can define whatever functions you want. for example, the function hlog (hybrid log) is included in the package. you can change easily write a function called mymedianpolish and use it instead of log2. -rafael On Thu, 5 Dec 2002, Sundaram, Shyam (NIH/CIT) wrote: > Hi: > I thought I would like to share a situation that I have been facing. It is > not a bug but something that happens in a specific condition. When you use > the "medianpolish" method in "expresso" and also use the "subtractmm" option > in the "pmcorrect" option you will find that the method will blow up. The > reason for this is sometimes the PM-MM will yield "negative" values and this > will cause the "NaN" values. > > I guess is it good to have the ability to choose the transformation function > to something other than log2 for negative values. > > Any suggestions/thoughts on this. > > Thanks > Shyam > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >