How to get DE genes without replicates using EdgeR

Question

[Aki]

Hi,

I am trying to perform DEG analysis without replicates while referencing EdgeR User's Guide(page23). I have 3 replicates but I want to compare each sample (cont3 vs sample1).

> library(edgeR)
> packageVersion("edgeR")[1] ‘3.30.3’
> count <- read.table("3vs1.txt", 
+                     sep = "\t", header = T, row.names = 1)
> head(count)
                        cont3 sample1
FBgn0000003_7SLRNA:CR32864   98.5      89
FBgn0000008_a               812.0     745
FBgn0000014_abd-A            32.0      38
FBgn0000015_Abd-B             2.0       2
FBgn0000017_Abl            2365.0    2469
FBgn0000018_abo             328.0     387
> dim(count)
[1] 17579     2

Then, I assigned values to

bcv <- 0.2

counts <- matrix (rnbinom(40, size = 1/bcv^2, mu=10), 20, 2)

y <- DGEList (counts=counts, group=1:2)

et <- exactTest(y, dispersion=bcv^2)

> bcv <- 0.2
> counts <- matrix(rnbinom(35158, size=1/bcv^2, mu=10), 17579, 2)
> y <- DGEList (counts=counts, group=1:2)
> et <- exactTest(y, dispersion=bcv^2)
> y
An object of class "DGEList"
$counts
    Sample1 Sample2
1       7       6
2      11      18
3       7      15
4      14      12
5      18      10
17574 more rows ...
$samples
    group lib.size norm.factors
Sample1     1   175534            1
Sample2     2   174455            1
> et
An object of class "DGEExact"
$table
      logFC   logCPM    PValue
1 -0.2094534 5.601973 1.0000000
2  0.7129906 6.559412 0.3761800
3  1.0947540 6.215569 0.2117326
4 -0.2114558 6.421354 0.8745086
5 -0.8312491 6.514097 0.2891758
17574 more rows ...
$comparison
[1] "1" "2"
$genes
NULL

I couldn't understand why gene name disappeared and indicated number appeared. Are these codes OK? Could someone please help me?

Answer 1

[Ido Tamir]

in the 2. part you called differential expression on a matrix without gene names as row names filled with random numbers

[Kevin Blighe]

As per my colleague, Ido, your y$counts object has no gene names, so, these will not be transmitted to the results table. You need to go back a few steps and set the gene names as rownames for y$counts

[Aki]

Thank you for your advice!! I googled how set the gene names as rownames, but I couldn't find. I'll continue to try searching for that.

[Kevin Blighe]

They (rownames) seem to be set in your original count variable; however, you then create a random counts matrix, counts (?), which has no rownames.

[Aki]

Thank you for everything you've done. This problem was all resolved.

bcv <- 0.2
counts <- matrix(rnbinom(35158, size=1/bcv^2, mu=10), 17579, 2)
rownames(counts) <- rownames(count)
y <- DGEList (counts=counts, group=1:2)
et <- exactTest(y, dispersion=bcv^2)

I am deeply grateful to you for your kindness.

[Kevin Blighe]

No problem, Aki.