Hello limma users. Can LIMMA fit a linear model to a Poisson distribution? My dataset is equivalent to an array but instead of intensities the genes have counts (it is sequence data). I would like to perform an eBayes() kind of analysis. However, is seems that this data has the attributes of a poisson distribution and (please correct me if I am wrong) fitting the linear model using lmFit() would not be suitable... So I wonder if there is another function that can fit a gene-wise linear model for this type of error distribution? sort of like glm(family=poisson) but for the whole array? And after this, would I be able to use eBayes()? I appreciate your insight. Thanks, Milena

Dear Milena, probably I am wrong here, since I don't know exactly what kind of data you have. But why don't you try Robinson & Smyth's packages for Tag abundances. I was going to point you to his website, but as of today there is a bioc package http://www.bioconductor.org/packages/2.3/bioc/html/edgeR.html. Also check out the papers by Robinson and Smyth. I guess this might better model your data. Best Wishes, bj?rn Milena Gongora wrote: > Hello limma users. > > Can LIMMA fit a linear model to a Poisson distribution? > > My dataset is equivalent to an array but instead of intensities the > genes have counts (it is sequence data). I would like to perform an > eBayes() kind of analysis. However, is seems that this data has the > attributes of a poisson distribution and (please correct me if I am > wrong) fitting the linear model using lmFit() would not be suitable... > > So I wonder if there is another function that can fit a gene-wise linear > model for this type of error distribution? sort of like > glm(family=poisson) but for the whole array? And after this, would I be > able to use eBayes()? > > I appreciate your insight. > > Thanks, > Milena > > _______________________________________________ > 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 > > This email was Anti Virus checked by the MPIMP-Golm Astaro Security Gateway. > > -- ---------------------------------------------- Bj?rn Usadel, PhD Max Planck Institute of Molecular Plant Physiology AG Integrative Carbon Biology Am Muehlenberg 1 14476 Potsdam-Golm Tel.: +49 331 5678153 email usadel at mpimp-golm.mpg.de ---------------------------------------------- This email was Anti Virus checked by the MPIMP-Golm Astaro Security Gateway.

Hi Milena. Short answer is no, limma cannot model count data. Gordon and I have developed a method that will operate on count data (such as nextgen sequencing) by doing moderation on the dispersion parameter of a negative binomial model: http://www.ncbi.nlm.nih.gov/pubmed/17881408 The R package will come out with Bioconductor 2.3, due for release on October 20th. You can get an advance copy at: http://www.bioconductor.org/packages/2.3/bioc/html/edgeR.html There are other groups working on similar moderated approaches but I haven't seen any results yet. Its a brand new package, so I'm keen to get some feedback. There is also a facility in the package for Poisson data. Cheers, Mark On 26/09/2008, at 3:39 PM, Milena Gongora wrote: > Hello limma users. > > Can LIMMA fit a linear model to a Poisson distribution? > > My dataset is equivalent to an array but instead of intensities the > genes have counts (it is sequence data). I would like to perform an > eBayes() kind of analysis. However, is seems that this data has the > attributes of a poisson distribution and (please correct me if I am > wrong) fitting the linear model using lmFit() would not be suitable... > > So I wonder if there is another function that can fit a gene-wise > linear model for this type of error distribution? sort of like > glm(family=poisson) but for the whole array? And after this, would I > be able to use eBayes()? > > I appreciate your insight. > > Thanks, > Milena > > _______________________________________________ > 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 ------------------------------ Mark Robinson Epigenetics Laboratory, Garvan Bioinformatics Division, WEHI e: m.robinson at garvan.org.au e: mrobinson at wehi.edu.au p: +61 (0)3 9345 2628 f: +61 (0)3 9347 0852