Hi, we have RNAseq data with some evident batch effect. We wont to correct this to have batch effect corrected counts to be then used in DESeq. The idea is to have all the subsequent analyses (e.g. PCA, hierarchical clustering, GSEA) that use DESeq normalized counts already compenseted for the batch effect. Is this possible? What is the best tool to remove batch effect on raw counts to obtain data to be then used in DESeq differential expression analysis? Thank you in advance for the help.

This is covered in the vignette. Including batch into the design is recommending for DE testing and removal from transformed counts is recommended for downstream analysis, see FAQs: https://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#why-after-vst-are-there-still-batches-in-the-pca-plot

The only way I know (not an official DESeq2 recommendation as per the vignette) to use corrected counts right away would be to use ComBat-Seq from the sva package which takes raw counts and returns batch-corrected raw counts ready for DESeq2.