EdgeR analysis with no replicates - how to run exact test with set dispersion value?
1
0
Entering edit mode
bertb • 0
@bertb-18667
Last seen 2.8 years ago
Canada

Hello,

I am trying to run a differential analysis with EdgeR for 2 RNAseq samples, with only one condition (1 treatment, 1 control), and I only have 1 replicate of each, which I know is far from ideal.

Having reviewed the user manual, I've decided to use option 2 for this scenario (on pg. 23), where I will choose a reasonable BCV, and use that to run the exact test.

Unfortunately, I'm having a difficult time understanding what commands to run in order to do this, despite the example detailed in the manual.

After normalizing my samples, my effective libraries are as follows:

> ynorm$samples
           group lib.size norm.factors
UWN_t2.bam     1  6411675    1.0307067
UMN_t2.bam     1  9047788    0.9702081

from reviewing the manual, I know I need to now set the BCV (with my chosen value of 0.2)

>bcv <- 0.2

but after that I don't know how to create a new DGEList which incorporates this new value, in order to run the exact test, since in the provided example there are some settings included that I cannot understand (size, mu, etc.).

Any advice would be greatly appreciated! Thank you,

edgeR • 3.1k views
ADD COMMENT
3
Entering edit mode
@james-w-macdonald-5106
Last seen 2 days ago
United States

The part you don't understand is just the creation of simulated data. You already have a DGEList called ynorm, so you don't need to generate any data. Instead you just go to the next step, which is fitting the exact test.

et <- exactTest(ynorm, dispersion = bcv^2)
topTags(et)
ADD COMMENT
0
Entering edit mode

I see - thank you for clarifying!

when I run the test as you suggested however, I receive the following error:

> et <- exactTest(ynorm, dispersion = bcv^2)
Error in exactTest(ynorm, dispersion = bcv^2) : 
  At least one element of given pair is not a group.
 Groups are: 1

Do I need to put each sample in a different group?

ADD REPLY
1
Entering edit mode

Yes. You should have included a group = 1:2 in your call to DGEList when you first made it.

ADD REPLY

Login before adding your answer.

Traffic: 823 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