Hi All,

I have a question related to the analysis of the microarray data. In the process of differential gene expression analysis, filtering low probe intensities is very crucial. The array chip I am analyzing is HuGene1.0 ST array. I am using oligo package to analyze and rma normalization. 

Can anyone recommend how to filter low probe intensities and how to derive a cut-off to know that these are the probes with low intensities?

Thanks a lot. 

Have a look at the IHW package, with overall variance (blind w.r.t. design factors) as a covariate. See also the accompanying paper.

Have a look at this workflow, which deals specifically with Affymetrix Gene type arrays and which covers filtering:

   https://f1000research.com/articles/5-1384/

If you're using the limma package to do the DE analysis with trend=TRUE, then the cutoff you choose for filtering is not in fact very crucial. You could look at plotSA(fit) after fitting the linear model to help judge whether more filtering could be done.