Hi guys, 

Bit of discussion question. When would you remove an outlier from an array dataset? 

I've run quite a few plots and analysis on my data including PCA, boxplots and lumi detectOutlier. I have focussed only on removing Controls that are outliers, since Cases may be statistically significant. 

I'm now returning to wonder when it's justifiable to even remove the controls? Are there any additional analyses you may recommend? 

I'd like to be careful with removing outliers, without jeopardising leaving in an obvious one. 

Cheerio

I remove outliers when there is something obviously amiss with that sample because something went wrong with the array or sample prep i.e. clustering gives some really weird outlier. Other than that, I leave them.