wyguo
commented
3 years ago Hi @canankolakoglu, The batch effect step doesn't directly remove the batch effect from the exprssion dataset. It will estimate a term of the effect from the data which can be incorporate in the linear regression of limma-voom step. That means the input data for differential expression analysis is still the data with batch effect. But a term has been added to the regression model to take into account the batch effects.
Hello,
Firstly, thank you for developing this useful RNAseq tool. I was reproducing 3D-RNAseq tool with R command line and at data preprocessing step I got confused. Throughout the this step we firstly remove the low expressed transcripts and then apply PCA. After that the batch effects are detected and corrected(?). And finally the normalization is applied.
But in R command line, at normalization step we are not using batch-effect-corrected data as input. Instead, we used low-expression-filtered data and normalize it. Shouldn't we use batch-effect-corrected data at normalization point?
Thank you