Hello group, We are working on microarray data analysis of 35 experiments with limma. We have used normexp technique as background correction, loess as within array nomalization and quantile as between array normalization. After doing quantile normalization I found large number of outliers (small circles) with box plot analysis.I would like to know whether the outliers are relevent in data analysis should they be removed or can be ignored? Also if they are to be removed can any one please let me know how to minimize these outliers? Thankfully A.Ashwin Department Of Biotechnology Manipal Life Science Center Karnataka INDIA

No advice can be given on the basis of the information you have supplied, and some of the concepts you mention ("minimize these outliers") are worrisome. >From a formal statistical perspective, testing for outliers can only be conducted in the context of a specific distributional model (see the monograph Outliers in Statistical Data by Barnett and Lewis for a host of algorithms; boxplot rules (there are various) are useful but informal. A primary function is to alert users to observations that "seem different" but that may very well be valid and compatible with all other non- outlying observations. In some contexts, the "outliers" are scientifically the most interesting observations, and one would not want to "minimize" them. In other contexts, the outliers arise through quality problems that can be identified in the data generation workflow, and data analysis needs to respond to the identified problems, not just to the observations that happen to be flagged as outliers. Several packages for reasoning about outlyingness in microarray data at various stages are present in Bioconductor: mdqc and arrayMvout ... take a look at these and at the associated references. On Thu, Apr 23, 2009 at 5:37 AM, Ashwin Kumar <ashwin.havoc@gmail.com>wrote: > Hello group, > > We are working on microarray data analysis of 35 experiments with > limma. We have used normexp technique as background correction, loess > as within array nomalization and quantile as between array > normalization. After doing quantile normalization I found large number > of outliers (small circles) with box plot analysis.I would like to > know whether the outliers are relevent in data analysis should they be > removed or can be ignored? Also if they are to be removed can any one > please let me know how to minimize these outliers? > > Thankfully > > A.Ashwin > Department Of Biotechnology > Manipal Life Science Center > Karnataka > INDIA > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > -- Vincent Carey, PhD Biostatistics, Channing Lab 617 525 2265 [[alternative HTML version deleted]]