Last seen 17 minutes ago
Republic of Ireland
Most importantly, you have not shown your DESeq2 code; however, I will answer as best as I can.
If, in your metadata that is supplied to DESeq2, '
control' is set as the reference level in your '
sampleCondition' column, then DESeq2 will use
control as the denominator for determining fold-changes. Then, you'll see a situation like this: http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#quick-start
In this case, a log2 value of +3 would indicate that the gene in question is more highly expressed (8 times on the linear scale) in '
condition' (or '
treated', for you) when compared to '
control' for you).
You can also manually set the order of comparison like this: http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#differential-expression-analysis, i.e..:
res <- results(dds, contrast=c("condition","treated","untreated"))
Here, the numerator is '
treated'; while the denominator is '
If in further doubt, just take a quick look via a box-and-whiskers:
boxplot(assay(vst)['VCAM1',] ~ colData(dds)[,'dex'])
Let's check the results table and do '
trt' versus '
res <- results(dds,
contrast = c('dex','trt','untrt'))
res <- lfcShrink(dds,
contrast = c('dex','trt','untrt'), res=res, type = 'normal')
log2 fold change (MAP): dex trt vs untrt
Wald test p-value: dex trt vs untrt
DataFrame with 1 row and 6 columns
baseMean log2FoldChange lfcSE stat pvalue padj
<numeric> <numeric> <numeric> <numeric> <numeric> <numeric>
VCAM1 509.995 -3.45198 0.177923 -19.307 4.69235e-83 4.72958e-80