Filtering 2-fold differentially expressed genes

Question

[vinayjrao]

Hi,

I am working with a microarray dataset, which is in the format -

Control      Control2      Treatment      Treatment2

The dataset has been log transformed. I want to know if there a way to filter the data such that I get genes whose fold difference is greater than 2 fold in Treatment as compared to Control.

For this purpose, I have already calculated the rowMeans() of each replicate to give

Control      Treatment

Thank you.

[Benn]

It is better to do DE analysis first, then after that if necessary you can include a Fold change filter. Try to use limma for your analysis, it has a very good manual.

Answer 1

[Nicolas Rosewick]

You should use limma to compute gene expression statistics (p-value, foldchange) and use this value to filter out your data.

https://bioconductor.org/packages/release/bioc/html/limma.html

Otherwise you can do something like this :

# control_samples # vector of control sample names
# treatment_samples # vector of treatment sample names
# M # gene expression matrix

fc <- rowMeans(M[,colnames(M) %in% control_samples]) / rowMeans(M[,colanames(M) %in% treatment_samples])

M.filtered <- M[ fc > 2,]