Hi,
Upon performing differential expression analysis for my RNA seq data set using DESeq I found that the log2 transformed differential expression values that I get from DESeq are very close to the RT-qPCR values but opposite in sign. I am new in the field and was concerned so as to why this would be happening? Is this something inherent to the way DESeq calculates the diff. values?
Thanks.
1 answer
when you perform a DESeq analysis, the program is comparing conditionB to conditionA, therefore the fold change will be a ratio of baseMeanB over baseMeanA. for example if you call DESeq like this:
nbinomTest( cds, "untreated", "treated" )
the fold change will be greater than 1 (and the log2 fold change positive), if the gene is more expressed in "treated". otherwise it will be lower than one (and the log2 fold change negative).
so you can simply change the sign to your log2 fold changes, or repeat the analysis changing the order of the conditions:
nbinomTest( cds, "treated", "untreated" )
Also.. like we can write the final differential expression analysis file with the log2, p-values etc. and save it, is there a way to obtain the file with the normalized values before we perform the DE analysis (right after normalization)? I tried a similar write command with the normalization file name instead of the DE analysis file but that gives an error.
cdsBlind<- estimateDispersions(cds,method="blind")
vsd<- varianceStabilizingTransformation(cdsBlind)
exprs(vsd) will be the normalized counts table
•
written
by
Ming Tommy Tang
Well that will give you the variance stabilized counts, which in general is not what you use in the DE.
To get the normalized counts table you can just use the function counts, setting the option normalized to TRUE:
counts(cds, normalized=TRUE)
(where cds is your CountDataSet object)
Hi,
This may be a weird question but, I have this normalized dataset that I want to analyze the same way as DESeq analyzes a data set normalized within it for differential expression.Is there a way to get DESeq to analyze it for differential expression (DE)? Or if not (as I read that it needs raw read data and normalized dataset cant be used) how can I use similar test of variance i.e. if I am not getting it wrong DESeq uses a chi square test can I replicate the same conditions(?) it uses elsewhere to simulate a similar DE analysis as by DESeq?
Log in to answer this question.
-
May I use Seurat package to analyze single cell RT-qPCR datasets generated on Biomark FluidigmN
written by francois-xavier.mauvais •
-
log2Fold Change as input for k-means analysis
written by concetta •
-
microRNA qPCR and RNAseq correlation
written by ehruan •
-
How to get the expression of sample combined the biological replicates in DESeq2?
written by rysterzhu •
-
gene expression analysis using RNA seq data
written by maryak •
-
How to get fitted per sample means and dispersion from DESeq2?
written by kjkjindal •
-
Analysis of microarray data using Genespring
written by jincyg38 •
-
What statistical test does DESeq use?
written by pmanga •
-
DESeq statistical test
written by pmanga •
-
Tcga Mirnaseq Differential Analysis: U-Test Or Edger-Like Approaches On Rpm Values
written by B. Arman Aksoy
Have you tried to swap the "reference" transcriptome? Like you did AxB, if you do BxA, maybe?
Hi,
Ya, I tried that and it would change the signs but was confused whether it was the right thing to do by swapping the reference as I wasn't sure if DESeq was comparing b to a or a to b.The following reply cleared that too. Thanks :)