Running DESeq2 without experimental conditions
Hi all! I tried running DESeq2 differential expression analysis on some RNA seq data without adding experimental conditions and DESeq2 actually outputs the results including log2 fold change and p-vals. Does anyone know what the results mean since there aren't any groups to compare with? Thank you!
• 991 views
•
link
updated
by
Ram
•
written
by
grsyhhsb •
•
written
by
grsyhhsb •
0 answers
No answers yet.
Log in to answer this question.
More posts like this
-
log2Fold Change as input for k-means analysis
written by concetta •
-
Manually calculating log2 fold change values from DESeq2 normalized counts
written by jayeshkumarsundaram •
-
Using bulk RNA-seq deconvolution as covariates in bulk RNA-seq differential expression analysis
written by telroyjatter
-
GWAS vs other bioinformatics methods
written by warriorgirl073095 •
-
RNA-seq compare FPKM's or fold change of different organisms
written by jehu •
-
Discrepancy in DESeq2 fold change
written by sophie.ehresmann •
-
Down-regulated genes showing up as red instead of green on R Pathview KEGG graph?
written by bionewbie •
-
Fold change > 2 and P-value<0.05 in micro-array differential analysis
written by Shicheng Guo
-
Differential expression with CQN normalized RNA-seq data
written by swkim •
-
Measuring Differential Expression Using Rna-Seq In Organisms Without A Genome Or Transcriptome?"
written by Senthilkumar •
Can you show code? It is to my knowledge not possible to run DESeq2 without a design, the only thing I could imagine is to run as intercept-only
~1but that makes no sense in a DE setting, see https://support.bioconductor.org/p/79209/Or do you mean running
results()without explicit name or contrast?Hello! Thanks for the reply. I did set it to run as intercept only, which I realized doesn't really mean anything. I was hoping to find genes that have the highest variability across samples. Upon doing some research, I think I should do Median Absolute Deviation analysis.
Do you have experimental replicates? Consider the LRT test in DESeq2 to find any differences between the groups. MAD does not mean a lot if you do not properly model variance as a function of the mean because variance is not equal across genes with different expression levels. The concepts of heteroscedasticity and overdispersion kick in here, which is why you should use expert software rather than custom approaches.