Hello,

I used DESeq2 to look for DE genes in a dataset of 15 samples with 3 conditions (C, A, and B).

C: control (5 replicates)

A: drug1 treatment (5 replicates)

B: drug2 treatment (5 replicates)

I found few or no gene with padj<0.1 for C vs. A, and C vs. B, respectively. Below is what I got. I'm wondering if the % of outlier and low counts look abnormal? Thanks for your help!

res1 <- results(dds, contrast=c("condition", "C", "A"))

summary(res1)

out of 15630 with nonzero total read count

adjusted p-value < 0.1

LFC > 0 (up)     : 0, 0%

LFC < 0 (down)   : 0, 0%

outliers [1]     : 2247, 14%

low counts [2]   : 0, 0%

(mean count < 0)

res2 <- results(dds, contrast=c("condition", "C", "B"))

summary(res2)

out of 15630 with nonzero total read count

adjusted p-value < 0.1

LFC > 0 (up)     : 3, 0.019%

LFC < 0 (down)   : 1, 0.0064%

outliers [1]     : 2247, 14%

low counts [2]   : 9931, 64%

(mean count < 1806.3)

sessionInfo()
R version 3.1.1 (2014-07-10)
Platform: x86_64-unknown-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] parallel  stats4    stats     graphics  grDevices utils     datasets
[8] methods   base     

other attached packages:
[1] DESeq2_1.6.3            RcppArmadillo_0.4.500.0 Rcpp_0.12.3            
[4] GenomicRanges_1.18.1    GenomeInfoDb_1.2.2      IRanges_2.0.1          
[7] S4Vectors_0.4.0         BiocGenerics_0.12.0    

First, a note: you should post the sessionInfo() for support site questions.

The filtering is typical for an experiment with no or very few DE genes. Although I guess you are not using a recent version of DESeq2, because filtering to a high mean count to recover a few genes no longer occurs with a more robust filtering procedure implemented a year ago.

The number of outliers is high, I would check the PCA plot for quality control of samples. Also the PCA plot gives a sense of whether the groups separate by gene expression.

The fact that there are no differentially expressed genes means that, either there are no differences in gene expression induced by the treatment, or that the variability is so high and the effects so small that 5 samples is not sufficient to detect them.