Between sample normalization without replicates

Question

[GR]

Hi All,

I have some RNA-Seq data without reps for 5 different conditions. I can not do any fancy statistics with this data. Just trying to get the idea if the genes of interest show stable expression in these conditions. What is the best way to normalize counts between samples?

I tried generating TPMs using salmon quant but it seems to do only within-sample normalization or multiple reps/libs concatenated together. Any suggestions for tools and methods for this?

Thanks!

[rpolicastro]

You can use the rlog/vst normalized counts from DESeq2 or TMM from edgeR.

[GR]

Thanks rpolicastro , can you please elaborate more why rlog/vst normalization should be used.

[jv]

To clarify, DESeq2 rlog and vst are transformation methods that DESeq will apply to counts that have been normalized for sequencing depth (i.e estimateSizeFactors).

For more details:

http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#data-transformations-and-visualization

http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#how-do-i-use-vst-or-rlog-data-for-differential-testing

[GR]

jv Thanks! I could not understand the purpose of using rlog or vst transformation from the deseq2 manual. I believe in my case normalizing counts by sequencing depth should be enough?