I'm working in the analysis of some RNAseq data with DeSeq2.

I have 2 conditions, treated and untreated, for two strains, mutant and wild type. With three replicates for each one, in total 12 samples.

I have done all the possible comparisons from the results function, specifying which experiments must be compared to define the numerator and denominator of the fold change relation.

Now I'm wondering if it is possible to get the standard deviation of the genes with significant differential expression across comparisons.

I have followed this manual https://www.bioconductor.org/packages/3.3/bioc/vignettes/DESeq2/inst/doc/DESeq2.pdf and it looks like it is possible to do it, but in some sort of separate way, it is possible to plot the expression levels of a specific gene through the conditions, section 1.5.2 and figure 2, and also its possible to plot the standard deviation of the transformed count data, section 2.1.5 and figure 4.

I'm not familiar with R, but in a naive guess, I think it is possible to select which data is going to be plotted from the result of the DESeq function on the count data (dds).

Could you please give some example or some insight on how I should do the calling or the filtering on the dds to get the standard deviation of the counts (transformed or not) for the significant genes, padj below 0.05?

Thank you

The standard deviation of the transformed count data in Figure 4 is calculated using the meanSdPlot function from the vsn package. This is just calculating the sample standard deviation, i.e.:

row.sd <- apply(m, 1, sd)

For a matrix, m.

However, I guess you want the within-group standard deviation of transformed data?

But I'm not sure what value this is, because the standard deviation of transformed counts is not used in the statistical testing, and with only 3 replicates per group, you would certainly only want to plot the normalized counts themselves, not a mean + sd summary.