What will lead DESeq2 shows no DEGs if workflow adopted is okay?
2
0
Entering edit mode
gmumuxi • 0
@gmumuxi-20104
Last seen 3.9 years ago

Hi,

ddsFullCountTable <- DESeqDataSetFromMatrix(countData = mouse_count, colData = mouse_pheno, design = ~ batch + age + gender + treatment)  

dds <- DESeq(ddsFullCountTable, test="LRT", reduced = ~ batch + age + gender)  

res_dds <- results(dds, contrast = c("treatment", "med", "ck"))

summary(res_dds) 
# out of 32102 with nonzero total read 
# adjusted p-value < 0.1
# LFC > 0 (up)       : 0, 0%
# LFC < 0 (down)     : 0, 0%
# outliers [1]       : 1269, 3.6%
# low counts [2]     : 0, 0%
# (mean count < 0)
# [1] see 'cooksCutoff' argument of ?results
# [2] see 'independentFiltering' argument of ?results

Here are 5 med mice and 7 check mice (so the ncols is 12 for countData) RNA-Seq data be used to do DE analysis, to find the DEGs affected by medicine, but seems no genes. This problem has confused me for a long time, I checked and tired many times, still haven't been solved. As far as I understand it, I guess:

  1. Too few samples (and too many covariates?)?
  2. The drug not statistically significant in this case?
  3. The code, I guess not?

Or other problems?

Thanks!

deseq2 • 450 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 2 hours ago
United States

Nothing wrong with the code.

There do seem to be a lot of outlier counts. Have you looked at plotPCA?

ADD COMMENT
0
Entering edit mode

Thanks for your reply! Since outliers here is 1269, it mean the "outlier" are genes, not samples. However I considered the outlier problem, it seems no obvious steps(function or tool) that to quantitative identification and remove them in DESeq2 wokflow?

ADD REPLY
0
Entering edit mode

Take a look at the vignette, these are removed by DESeq2. It explains in detail.

ADD REPLY
0
Entering edit mode
arfranco ▴ 130
@arfranco-8341
Last seen 20 hours ago
European Union

Run a plotPCA like Michael recommends you

I had an experiment with 2 conditions and 3 biological replicates. By running a PCA analysis, I noticed that two samples were exchanged (a control was actually a treated sample, and viceversa), and this was enough for not getting a single DE gene

ADD COMMENT

Login before adding your answer.

Traffic: 360 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6