Dear Sir, I am in need of your help to decide which test would be an appropraite in this case. I am involved in the data analysis of the out put from Illumina. I have the following experimental design Array A : TBVII Treat 01 Array B : TBVII Treat 02 Array c : TBVII Treat 03 (they are biological replicates) Array D : SSC15 -01 (just a sample without replicate) Array E : TBVII control 01 Array F: TBVII control 02 (They are technical replicate) I understand that this is a Bad experimental design but a group wants to do have some preliminary information to have from this. Kindly help me out to decide what sort of statistical test I can apply for the anlaysis. There is sample without any replicate. It would be of great help since I am only a beginner for this type of analysis. Thanking You, antony -- John Antonydas Gaspar, Phd Student Institute of Neurophysiology University of Cologne Robert-Koch-Str. 39 50931 Cologne/Germany Tel: 004922125918042 Handy: 004917683142627

On Thu, Nov 20, 2008 at 5:37 AM, John Antonydas Gaspar <gasparj@uni- koeln.de=""> wrote: > Dear Sir, > > I am in need of your help to decide which test would be an appropraite in > this > case. > > I am involved in the data analysis of the out put from Illumina. > I have the following experimental design > > Array A : TBVII Treat 01 > Array B : TBVII Treat 02 > Array c : TBVII Treat 03 > (they are biological replicates) > Array D : SSC15 -01 > (just a sample without replicate) > Array E : TBVII control 01 > Array F: TBVII control 02 > (They are technical replicate) > > I understand that this is a Bad experimental design but a group wants to do > have > some preliminary information to have from this. > > Kindly help me out to decide what sort of statistical test I can apply for > the > anlaysis. There is sample without any replicate. It would be of great help > since I am only a beginner for this type of analysis. > I would simply make a heatmap of the top 200 or so varied genes or those with the largest fold changes. There are not meaningful statistics that can be done with an n of 1. Sean [[alternative HTML version deleted]]

Actually, you can do a statistical test if you are willing to assume that the variance does not change with treatment (which is the usual ANOVA assumption), because the first set of arrays includes biological replication. However, I have not offered advice on this design because I am not familiar with Illumina arrays. If they are like 1-channel arrays, you can use limma or maanova. There are 3 treatments, and the technical replicate is a block. This is highly unbalanced, and you analysis would probably be simpler if you take the "better" of the 2 technical replicates, where "better" refers to array quality. Then you just have an unbalanced 1-way ANOVA and analysis in limma or maanova should be straight-forward. If they are not like 1-channel arrays, then I defer to those who are familiar with this platform. --Naomi At 10:45 AM 11/20/2008, Sean Davis wrote: >On Thu, Nov 20, 2008 at 5:37 AM, John Antonydas Gaspar <gasparj at="" uni-koeln.de=""> > wrote: > > > Dear Sir, > > > > I am in need of your help to decide which test would be an appropraite in > > this > > case. > > > > I am involved in the data analysis of the out put from Illumina. > > I have the following experimental design > > > > Array A : TBVII Treat 01 > > Array B : TBVII Treat 02 > > Array c : TBVII Treat 03 > > (they are biological replicates) > > Array D : SSC15 -01 > > (just a sample without replicate) > > Array E : TBVII control 01 > > Array F: TBVII control 02 > > (They are technical replicate) > > > > I understand that this is a Bad experimental design but a group wants to do > > have > > some preliminary information to have from this. > > > > Kindly help me out to decide what sort of statistical test I can apply for > > the > > anlaysis. There is sample without any replicate. It would be of great help > > since I am only a beginner for this type of analysis. > > > >I would simply make a heatmap of the top 200 or so varied genes or those >with the largest fold changes. There are not meaningful statistics that can >be done with an n of 1. > >Sean > > [[alternative HTML version deleted]] > >_______________________________________________ >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 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