Hi there, I am trying to automate an RNA-seq protocol based on the normalisation function RUVs which determines k factors of unwanted variation in your dataset that are then fitted in the model.matrix that you provide to edgeR in a way like this design <- model.matrix(~0+treatments + W_1 + W_2 + W_3 + W_4 + W_5, data=pData(set)). This example has k=5 factors of unwanted variation (stored in my pData(set) object). If with another dataset I had three factors of unwanted variation, then the code would be design <- model.matrix(~0+treatments + W_1 + W_2 + W_3, data=pData(set)). How can I have a "generalised" version of the code that accounts for all the unwanted factors of variation in my pData(set) object without the need to manually type them into the formula?

Thanks in advance Luca

If you already know that you will be using 'treatments' and W1 ... WN, you can just extract them from the column names of your pData object:

ws <- grep("W_", colnames(pData(set)), value = TRUE)
form <- as.formula(paste("~ 0 + treatments +", paste(ws, collapse = " + ")))
design <- model.matrix(form, pData(set))