Dear,

I would like to perform a differential expression analysis from TCGA data. I have found the file "EBPlusPlusAdjustPANCANIlluminaHiSeqRNASeqV2.geneExp.tsv" from their publication website https://gdc.cancer.gov/about-data/publications/PanCan-CellOfOrigin . I understand that this was done using RSEM quantification and has already been upper-quantile normalised, which should not be possible to use standard packages such as DESeq2 and limma. My question is what is the best way to do a differential expression analysis from this file?

Thank you.

Dear KCLiv,

if i have understood correctly you have "normalized estimated counts from RSEM" - i have faced previously a similar situation and you can check one of my previous posts with suggested solutions-briefly:

https://support.bioconductor.org/p/91054/#95981

1) Continue with the limma-trend pipeline

2) Use an alternative solution with the R package tximport, starting with the transcript abundance estimates and use other bioconductor packages for DE expression like edgeR or DESeq2 http://bioconductor.org/packages/release/bioc/vignettes/tximport/inst/doc/tximport.html#rsem

But above all, i would suggest to check in detail if extra transformations have been performed, such as log2 or other pre-processing steps.

Best,

Efstathios