Dear there,

From the TCGAbiolinks paper (https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1006701), I see it need scaling Recount2 data because it was sequenced using Rail-RNA. Here is the code they used as below for downstream differential expression analysis (DEA) using "TCGAanalyze_DEA". They also used "TCGAanalyze_Normalization" and "TCGAanalyze_Filtering" to do the normalization and filtering. Since DESeq2 takes the raw read counts as input, I am wondering do I need any of these 3 (scaling, normalization, and filtering) if I used "TCGAquery_recount2" to download GTEx and TCGA data? Thanks a lot!

Xiaofei

#####Preparing/scaling Recount2 data because it was sequenced using Rail-RNA

eset.gtex<-assays(scale_counts(ucs.recount.gtex$GTEX_uterus, round = TRUE))$counts
eset.tcga<-assays(scale_counts(ucs.recount.tcga$TCGA_uterus, round = TRUE))$counts

recount has a function scale_counts to create a count matrix appropriate for running DESeq2, edgeR, limma-voom or other count based packages. From their vignette, in the code just before running DESeq2:

## Scale counts by taking into account the total coverage per sample
rse <- scale_counts(rse_gene)

If that function has been run, then you are set.