Dear colleagues, Hi, I would like to ask a question related to creating design matrix in limma. Statistics is not my strong subject so I apologise in advance if this is a silly question. I have a number of arrays with 3 different treatments and I can generate a design matrix using modelMatrix. > modelMatrix(targets, ref="Pool") Found unique target names: one Pool three two one three two samp01 1 0 0 samp02 0 0 1 samp03 1 0 0 ...etc But there are other nuisance factors, for example the arrays were hybridised in 3 different labs, and the animals are not all from the same line. I am only interested in comparing the treatment effect but in the analysis I would like to account for the nuisance factors. I have been reading the limma user guide and I guess this is a not really factorial design, correct? Any idea on how to make this matrix? Regards, Alex ------------------------------------ Alex Lam PhD student Department of Genetics and Genomics Roslin Institute (Edinburgh) Roslin Midlothian EH25 9PS Phone +44 131 5274471 Web http://www.roslin.ac.uk

Are these 2-channel arrays? With 1 sample per treatment you cannot do statistical testing, even without the nuisance factors. =--Naomi At 09:43 AM 2/10/2006, alex lam (RI) wrote: >Dear colleagues, >Hi, I would like to ask a question related to creating design matrix in >limma. Statistics is not my strong subject so I apologise in advance if >this is a silly question. I have a number of arrays with 3 different >treatments and I can generate a design matrix using modelMatrix. > > > modelMatrix(targets, ref="Pool") >Found unique target names: > one Pool three two > one three two >samp01 1 0 0 >samp02 0 0 1 >samp03 1 0 0 >...etc > >But there are other nuisance factors, for example the arrays were >hybridised in 3 different labs, and the animals are not all from the >same line. I am only interested in comparing the treatment effect but in >the analysis I would like to account for the nuisance factors. I have >been reading the limma user guide and I guess this is a not really >factorial design, correct? Any idea on how to make this matrix? > >Regards, >Alex > >------------------------------------ >Alex Lam >PhD student >Department of Genetics and Genomics >Roslin Institute (Edinburgh) >Roslin >Midlothian EH25 9PS > >Phone +44 131 5274471 >Web http://www.roslin.ac.uk > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor Naomi S. Altman 814-865-3791 (voice) Associate Professor Dept. of Statistics 814-863-7114 (fax) Penn State University 814-865-1348 (Statistics) University Park, PA 16802-2111

There is quite a bit of discussion on this list about combining data across labs and dealing with other nuisance factors. There are issues in normalization, lab effects, etc. I think you will learn a lot from reading these - but you will likely not find a complete satisfactory answer. --Naomi At 10:34 AM 2/10/2006, alex lam (RI) wrote: >Dear Naomi, >They are 2-channel arrays but there are many samples. 115 arrays, in >fact. I just didn't paste in the whole matrix. And the treatment is >actually a SNP genotype, hence 3 treatments. I problem is that there >are other factors - line and lab that I want to account for as >background nuisance effect, but I don't want to compare 1 genotype >from 1 lab with another genotype from another lab. I am interested >in comparing the three treatment group. > >Thanks, > >Alex C. Lam >PhD student >Dept. of Genetics and Genomics >Roslin Institute, Edinburgh >EH25 9PS >UK > > > >-----Original Message----- >From: Naomi Altman [mailto:naomi at stat.psu.edu] >Sent: Fri 2/10/2006 3:05 PM >To: alex lam (RI); bioconductor at stat.math.ethz.ch >Subject: Re: [BioC] Limma question > >Are these 2-channel arrays? With 1 sample per treatment you cannot >do statistical testing, even without the nuisance factors. > >=--Naomi > >At 09:43 AM 2/10/2006, alex lam (RI) wrote: > > >Dear colleagues, > >Hi, I would like to ask a question related to creating design matrix in > >limma. Statistics is not my strong subject so I apologise in advance if > >this is a silly question. I have a number of arrays with 3 different > >treatments and I can generate a design matrix using modelMatrix. > > > > > modelMatrix(targets, ref="Pool") > >Found unique target names: > > one Pool three two > > one three two > >samp01 1 0 0 > >samp02 0 0 1 > >samp03 1 0 0 > >...etc > > > >But there are other nuisance factors, for example the arrays were > >hybridised in 3 different labs, and the animals are not all from the > >same line. I am only interested in comparing the treatment effect but in > >the analysis I would like to account for the nuisance factors. I have > >been reading the limma user guide and I guess this is a not really > >factorial design, correct? Any idea on how to make this matrix? > > > >Regards, > >Alex > > > >------------------------------------ > >Alex Lam > >PhD student > >Department of Genetics and Genomics > >Roslin Institute (Edinburgh) > >Roslin > >Midlothian EH25 9PS > > > >Phone +44 131 5274471 > >Web http://www.roslin.ac.uk > > > >_______________________________________________ > >Bioconductor mailing list > >Bioconductor at stat.math.ethz.ch > >https://stat.ethz.ch/mailman/listinfo/bioconductor > >Naomi S. Altman 814-865-3791 (voice) >Associate Professor >Dept. of Statistics 814-863-7114 (fax) >Penn State University 814-865-1348 (Statistics) >University Park, PA 16802-2111 Naomi S. Altman 814-865-3791 (voice) Associate Professor Dept. of Statistics 814-863-7114 (fax) Penn State University 814-865-1348 (Statistics) University Park, PA 16802-2111

Dear Naomi, They are 2-channel arrays but there are many samples. 115 arrays, in fact. I just didn't paste in the whole matrix. And the treatment is actually a SNP genotype, hence 3 treatments. I problem is that there are other factors - line and lab that I want to account for as background nuisance effect, but I don't want to compare 1 genotype from 1 lab with another genotype from another lab. I am interested in comparing the three treatment group. Thanks, Alex C. Lam PhD student Dept. of Genetics and Genomics Roslin Institute, Edinburgh EH25 9PS UK -----Original Message----- From: Naomi Altman [mailto:naomi@stat.psu.edu] Sent: Fri 2/10/2006 3:05 PM To: alex lam (RI); bioconductor at stat.math.ethz.ch Subject: Re: [BioC] Limma question Are these 2-channel arrays? With 1 sample per treatment you cannot do statistical testing, even without the nuisance factors. =--Naomi At 09:43 AM 2/10/2006, alex lam (RI) wrote: >Dear colleagues, >Hi, I would like to ask a question related to creating design matrix in >limma. Statistics is not my strong subject so I apologise in advance if >this is a silly question. I have a number of arrays with 3 different >treatments and I can generate a design matrix using modelMatrix. > > > modelMatrix(targets, ref="Pool") >Found unique target names: > one Pool three two > one three two >samp01 1 0 0 >samp02 0 0 1 >samp03 1 0 0 >...etc > >But there are other nuisance factors, for example the arrays were >hybridised in 3 different labs, and the animals are not all from the >same line. I am only interested in comparing the treatment effect but in >the analysis I would like to account for the nuisance factors. I have >been reading the limma user guide and I guess this is a not really >factorial design, correct? Any idea on how to make this matrix? > >Regards, >Alex > >------------------------------------ >Alex Lam >PhD student >Department of Genetics and Genomics >Roslin Institute (Edinburgh) >Roslin >Midlothian EH25 9PS > >Phone +44 131 5274471 >Web http://www.roslin.ac.uk > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor Naomi S. Altman 814-865-3791 (voice) Associate Professor Dept. of Statistics 814-863-7114 (fax) Penn State University 814-865-1348 (Statistics) University Park, PA 16802-2111