Dear James, I completely understand the importance of me learning on my own. But, the thing here is this data is not my own experiment. I have been given these 4 files as a test for a job interview. I do not have any other detail regarding it. I just have these 4 cel files. I know the first 3 are wild types or control and the other one is a tumor. I have to find the differential expression of tumor as compared to control. I have been reading up on the case studies as well as limma user manual. I have understood that my case is the two groups :Affymetrix case as according to section 8.5 of limma user manual. But, as per there manual , I am trying to create my design matrix and still I am not able to get it. I tried doing the contrast matrix also but I am still not getting it. I thought a lot about it and now I used the following command : design <- model.matrix(~factor(rep(1:2, c(3,1)))) fit <- lmFit(eset, design) fit2 <- eBayes(fit) topTable(fit2, 1) ID logFC AveExpr t P.Value adj.P.Val 973 1416642_PM_a_at 13.64906 13.60996 208.6925 6.608690e-09 1.436053e-06 8941 1424635_PM_at 13.55811 13.52669 206.8301 6.839348e-09 1.436053e-06 22282 1437976_PM_x_at 13.41281 13.41262 205.0409 7.070582e-09 1.436053e-06 955 1416624_PM_a_at 13.61291 13.61261 204.7329 7.111384e-09 1.436053e-06 32409 1448109_PM_a_at 13.35854 13.39007 203.3663 7.296015e-09 1.436053e-06 548 1416217_PM_a_at 13.31925 13.33522 202.1300 7.468282e-09 1.436053e-06 21301 1436995_PM_a_at 13.33443 13.35885 202.0929 7.473534e-09 1.436053e-06 33736 1449436_PM_s_at 13.36569 13.33920 201.5598 7.549461e-09 1.436053e-06 1748 1417417_PM_a_at 13.36793 13.26406 201.5213 7.554988e-09 1.436053e-06 8069 1423763_PM_x_at 13.20019 13.21718 200.1702 7.752016e-09 1.436053e-06 B 973 9.389064 8941 9.383131 22282 9.377292 955 9.376273 32409 9.371700 548 9.367491 21301 9.367364 33736 9.365526 1748 9.365393 8069 9.360674 > topTable(fit2, 2) ID logFC AveExpr t P.Value adj.P.Val 44020 1459725_PM_s_at 9.002733 5.546952 66.37302 5.294023e-07 0.0064537 21237 1436931_PM_at -7.605935 9.492585 -53.42400 1.213995e-06 0.0064537 5418 1421112_PM_at -6.853626 7.778874 -51.89407 1.356622e-06 0.0064537 39116 1454821_PM_at -6.814400 9.007994 -48.96033 1.694613e-06 0.0064537 8985 1424679_PM_at 6.579054 5.260500 48.16929 1.803474e-06 0.0064537 3964 1419633_PM_at 6.547493 4.696412 48.04883 1.820817e-06 0.0064537 8706 1424400_PM_a_at -8.609042 9.895516 -47.03876 1.974842e-06 0.0064537 7235 1422929_PM_s_at 6.579742 6.075454 46.85107 2.005251e-06 0.0064537 34850 1450555_PM_at 7.541370 4.892047 46.41624 2.077999e-06 0.0064537 1790 1417459_PM_at 6.722704 5.163969 45.95859 2.158198e-06 0.0064537 B 44020 5.930831 21237 5.650912 5418 5.606440 39116 5.511995 8985 5.484246 3964 5.479928 8706 5.442737 7235 5.435626 34850 5.418903 1790 5.400918 Is this approach correct?. I know I am asking a bit too much but this is the first time I am trying it and so want to be sure about it. Also , if possible can you tell me the difference in both the topTables. Which of the two tables are tbe difference between Tumor and control. What I understand is that co-ef = 2 is the one of my interest because it gives the differential expression in tumor. I am really sorry for so many questions and I hope this will clear up the situation a little bit. Thanks, Himanshu Sharma. > Date: Wed, 18 Apr 2012 14:50:15 -0400 > From: jmacdon@uw.edu > To: hsharm03@students.poly.edu > CC: bioconductor@r-project.org > Subject: Re: [BioC] Affy PhenoData > > Hi Himanshu, > > On 4/18/2012 12:27 PM, hsharm03@students.poly.edu wrote: > > Dear all, > > I have 4 samples of HT 430mgpm array plate. Three of them are replicates of wild types and 1 is a tumor condition with no replicates. But I do not have any other information regarding how was the experiment conducted and I am not able to figure out how to create a phenodata of the same. When I create the expression set of this data using rma function of affy library I can see the names of the samples as they were and sample numbers namely 1,2,3,4. What I understand is it takes alll the four samples as different and when I do the differential expression analysis using limma , I try to create the model.matrix using the following command : > > > > design<- model.matrix(~sample, pData(eset)) > > > > But what I understand is that the sample that are present in the data it is taking 1 condition each of 4 samples. Am I understanding it correctly?. > > Yes, you are understanding it correctly. But this leads me to a separate > point, below. > > > If so what should I be doing to get differential expression of genes in tumor as compare to the 3 wild type replicates that I have . > > The simple answer is that you should use the correct input to > model.matrix(), designed for your experiment. I realize that is a vague > and wholly unsatisfying answer, but we have arrived at the point for you > that occurs for all long term R users, when they either decide to figure > stuff out themselves or they become disillusioned and give up. > > It would be simple for me to tell you exactly what you need to do for > this step in your analysis. And then answer the next question, and the > one after that. But that helps no one. > > If you are really going to analyze your own data (not recommended IMO, > but that's the beauty of Open Source software - we get both the rope and > the tree, and are free to hang ourselves) you will have to learn how to > figure out both what you should be doing, and how to do it. > > So the best advice I can offer is to recommend that you find a local > statistician to help with your analysis. Barring that, you should > closely read the 'limma User's guide', probably the 'Introduction to R', > certainly look at ?formula, ?model.matrix, and ?factor. The Bioconductor > Case Studies contain many useful examples. There are also any number of > presentations given over the years that you could find via the BioC > website, or with good google skills. > > Best, > > Jim > > > > > > I am very new to this field and so I am not sure how to proceed > > Any help will be much appreciated. > > Thanks , > > Himanshu Sharma. > > > > [[alternative HTML version deleted]] > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor@r-project.org > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor > > -- > James W. MacDonald, M.S. > Biostatistician > University of Washington > Environmental and Occupational Health Sciences > 4225 Roosevelt Way NE, # 100 > Seattle WA 98105-6099 > [[alternative HTML version deleted]]