When I get expression value from deseq, the mean for all samples are not on the same level. Should I do quantile normalization on these data?

I get the expression data as follow:

ddsHTSeq<-DESeqDataSetFromHTSeqCount(sampleTable=sampleTable, directory=dat.dir, design=~1)
dds<-DESeq(ddsHTSeq)
fil <- rowSums(counts(dds)>1) >= 5
dds <- dds[fil,]

rld <- rlog(dds, blind=TRUE)
vsd <- varianceStabilizingTransformation(dds, blind=TRUE)

sessionInfo()
R version 3.1.2 (2014-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] parallel  stats4    datasets  utils     stats     graphics  grDevices
[8] methods   base     

other attached packages:
 [1] DESeq2_1.6.2                RcppArmadillo_0.4.550.1.0  
 [3] Rcpp_0.11.3                 GenomicRanges_1.18.3       
 [5] GenomeInfoDb_1.2.3          IRanges_2.0.0              
 [7] S4Vectors_0.4.0             BiocGenerics_0.12.1        
 [9] rgl_0.93.1098               scatterplot3d_0.3-35       
[11] stringr_0.6.2               nutshell_2.0               
[13] nutshell.audioscrobbler_1.0 nutshell.bbdb_1.0          
[15] faraway_1.0.6               MASS_7.3-35                

loaded via a namespace (and not attached):
 [1] acepack_1.3-3.3      annotate_1.44.0      AnnotationDbi_1.28.1
 [4] base64enc_0.1-2      BatchJobs_1.5        BBmisc_1.8          
 [7] Biobase_2.26.0       BiocParallel_1.0.0   brew_1.0-6          
[10] checkmate_1.5.0      cluster_1.15.3       codetools_0.2-9     
[13] colorspace_1.2-4     DBI_0.3.1            digest_0.6.4        
[16] fail_1.2             foreach_1.4.2        foreign_0.8-61      
[19] Formula_1.1-2        genefilter_1.48.1    geneplotter_1.44.0  
[22] ggplot2_1.0.0        grid_3.1.2           gtable_0.1.2        
[25] Hmisc_3.14-6         iterators_1.0.7      lattice_0.20-29     
[28] latticeExtra_0.6-26  locfit_1.5-9.1       munsell_0.4.2       
[31] nnet_7.3-8           plyr_1.8.1           proto_0.3-10        
[34] RColorBrewer_1.0-5   reshape2_1.4         rpart_4.1-8         
[37] RSQLite_1.0.0        scales_0.2.4         sendmailR_1.2-1     
[40] splines_3.1.2        survival_2.37-7      tools_3.1.2         
[43] XML_3.98-1.1         xtable_1.7-4         XVector_0.6.0

Hi Peter,

This is fine if the mean (or the median, the dark line in the boxplots) is not exactly the same number (equal medians is guaranteed by quantile normalization), as long as the boxes of a boxplot (the middle 50% of values) are mostly overlapping, then you can be assured that the library size has been normalized out, and that the values are on the same scale.