Dear All,

I would like to ask about using the voom-limma workflow on RNAseq data. Usually, when I run the workflow, the samples contain similar number of reads (translating into a similar number of counts between the samples); however, I have received data for 60 samples, each having 4-5.5M reads, with 2 samples having approximately 26M reads each.

My question is whether the voom-limma workflow will be able to "deal" with such a situation of such different amounts of reads or would this skew the results? If the latter is correct, what would you suggest to do to allow using these 2 samples?

Thank you very much!

Following is a general code I'm using for preparing the data for the differential limma analysis (originally taken from the limma guide). I am not providing a code specific to this situation with 60 samples, since I see my question as a more general one.

Thank you! Any advice will be greatly appreciated!

Regards,

Anna

library(limma)
library(edgeR)

dge <- DGEList(counts=dataset) # dataset - a matrix of genes x samples, containing counts.
design<-model.matrix(~0+factor(c(rep(1,3),rep(2,3),rep(3,3),rep(4,3))));  # an example design
keep <- filterByExpr(dge, design)
dge <- dge[keep,,keep.lib.sizes=FALSE]
dge <- calcNormFactors(dge)
va<-voom(dge,design,plot=TRUE);

sessionInfo( )

R version 3.5.2 (2018-12-20)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 17763)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.1252  LC_CTYPE=English_Canada.1252   
[3] LC_MONETARY=English_Canada.1252 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] edgeR_3.24.3 limma_3.38.3

loaded via a namespace (and not attached):
[1] compiler_3.5.2  Rcpp_1.0.1      grid_3.5.2      locfit_1.5-9.1 
[5] lattice_0.20-38

Yes, voom handles different library sizes, in fact doing so was one of the primary aims of the method. The original voom paper (Law et al 2014) benchmarked voom on simulated datesets with library sizes varying 10-fold, far more different than are the library sizes of your samples.