Remove unknown batch effect in microarrays data

Question

[antonio.dos.hermanas]

Hi everyone!

Currently, I am working with microarray raw data. After background correction, normalization and summarization I realized that my data had a huge batch effect when I used a principal component analysis (PCA).

I know how to remove batch effects when they are known, however, in this case, I have no idea what effect is.

I have tried to use packages in R such as ruv or sva, but they don't return the corrected matrix. My problem is that I need this corrected matrix for downstream analysis.

What can I do?

[boxate1618]

what kind of array is it ?

[antonio.dos.hermanas]

It's a microarray data, concretely, an Agilent microarray data.

Answer 1

[Papyrus]

The removeBatchEffect function in limma allows you to regress-out continuous covariates (it can also "protect" your design/comparison on the way). Also look at other posts like this one, this one, etc. (remember that for testing it is best to include the covariates in the model with the uncorrected data).