the difference between RNA-seq FPKM and Affymetrix array on gene expression

Question

[zhao03]

I build a model based on RNA-seq FPKM value , can i directly apply the Affymetrix array gene expression value to validate the model? how to transform Affymetrix array data into FPKM value? thank you very much!

Answer 1

[dsull]

FPKM and microarray values aren't directly comparable. Also, I wouldn't recommend using FPKMs for RNA-seq in the first place.

If you're doing some rank-based correlation model (e.g. calculating Spearman's rho between your RNA-seq expression data and your microarray expression data), then it might work. But if you're actually trying to make FPKM values from microarray probe intensity values, it's not possible. If you're training a machine learning model on RNA-seq datasets and then testing the model on microarray data, there really is no perfect way to do it -- you can try quantile normalization of your datasets (to make the distributions similar) but even that won't remove a lot of the differences between microarray and RNA-seq data.