How to Merge RNA Replicates

Question

[cthangav]

I am following the manual for a program called TimeReg that says "If there are multiple replicates, merge them to get one expression profile. For gene expression data, you may use the average expression (FPKM or TPM) of the replicates."

I have two replicates for each RNA sample I would need to get averages for.

I have my expression data in counts:

geneID  geneLength  counts 
ENSG00000121410 4006    1 
ENSG00000268895 2793    0 
ENSG00000148584 9603    0
ENSG00000175899 6384    0
ENSG00000166535 7303    0
ENSG00000256661 452 0
ENSG00000256904 386 0
ENSG00000256069 4657    0
ENSG00000128274 2943    20

Which I converted to FPKMs with EdgeR.

What is the best way to take the average of each pair of replicates like it says in the manual? I was thinking of taking the average of the counts and using EdgeR on the averaged counts to get FPKMs.

Answer 1

[Gordon Smyth]

If you're using edgeR, the correct (best) way to average biological replicates is to use the cpmByGroup or rpkmByGroup functions. These functions take into account differences in sequencing depth between the samples and the amount of biological variation.

If your replicates are purely technical (just re-sequencing of the same RNA samples) then use sumTechReps.