Dear all, I have some basic questions about two-color array. For two-color array with common reference method, one color is for interested sample and the other is for common reference. Generally, the output is the ratio between interested sample and common reference sample after a Lowess method. 1) Can I simply regard these ratios as gene expression values (used in one-color array) and directly use them to do some analysis since a common reference is used? Or I have to calculate the expression value of interested sample from raw data (onlyl use one channel)? 2) Should I do some inter-chip normalization such as Quantile method used in one-colar array? 3) Should I check the batch effect? Is there popular or standard method to check and alleviate bache effect? Thank you very much. Feng Tian Boston University [[alternative HTML version deleted]]

On Tue, Nov 6, 2012 at 12:39 AM, Feng Tian <fengtian@bu.edu> wrote: > Dear all, > I have some basic questions about two-color array. > For two-color array with common reference method, one color is for > interested sample and the other is for common reference. Generally, the > output is the ratio between interested sample and common reference sample > after a Lowess method. > 1) Can I simply regard these ratios as gene expression values (used in > one-color array) and directly use them to do some analysis since a common > reference is used? Yes the log ratios can generally be treated the same as values from single channel arrays. > Or I have to calculate the expression value of > interested sample from raw data (onlyl use one channel)? > 2) Should I do some inter-chip normalization such as Quantile method used > in one-colar array? > Yes, normalizing the log ratios is usually necessary. Check manufacturers guidelines / previously published papers using the same platform to find out what method is most suitable, but quantiles is probably fine. > 3) Should I check the batch effect? Is there popular or standard method to > check and alleviate bache effect? > You could probably check for batch effect using PCA or a similar clustering algorithm. Limma implements functionality to correct for batch effect (if there is evidence of this). Paul. > Thank you very much. > > Feng Tian > > Boston University > > [[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 > -- Dr. Paul Geeleher School of Mathematics, Statistics and Applied Mathematics National University of Ireland Galway Ireland -- www.bioinformaticstutorials.com [[alternative HTML version deleted]]