Which is the best way for GSVA on TCGA dataset

Question

[Yang Shi]

Dear Communities,

My question is which kind of data from TCGA is the best way for GSVA analysis. This post (GSVA R packages) demonstrated the best way is based on row counts data. However, FPKM, TPM, RPKM are also acceptable for GSVA (kcdf = 'Gaussian'). So I wonder why row count is the best way for GSVA. Any suggestions would be appreciated!

Answer 1

[Yang Shi]

Furthermore, if there is no control sample, e.g. ACC (TCGA), how to get the normalised count from row count?