edgeR and variance
2
0
Entering edit mode
@nooriaalwathiqui-8916
Last seen 8.5 years ago
United States

Is it possible to have edgeR output contigwise variance? I'm looking to determine within and between group variance. 

edger • 1.3k views
ADD COMMENT
0
Entering edit mode

What is "contigwise variance"?

ADD REPLY
0
Entering edit mode

Has anyone solved this problem!? I have the same exact question: looking to determine within and between group variance for each contig.

ADD REPLY
2
Entering edit mode
@gordon-smyth
Last seen 2 hours ago
WEHI, Melbourne, Australia

Honestly, I don't think that you're looking for the right thing. Looking for with and between variances from edgeR is a bit like looking for propellers on a jet plane.

First, variances are only appropriate for normally distributed observations. Variances are not an appropriate measure of variability for RNA-seq counts.

Second, within and between group variances are only a means to an end, not an end in themselves. What you ultimately need to do is assess differential expression, and edgeR provides a wealth of ways to do that. It would be better to use edgeR on its own terms, rather than trying to repeat a traditional analysis that is not appropriate for this type of data.

ADD COMMENT
1
Entering edit mode
Aaron Lun ★ 28k
@alun
Last seen 1 day ago
The city by the bay

edgeR will analyze whatever counts you give it. Most typically these are counts for a gene across all samples, but they can also be counts for genomic regions, transcripts, etc. Supplying counts for a contig will work just fine; in which case, the tagwise NB dispersion corresponds to the "within group variance" for each contig. I presume you got the "between group variance" concept from ANOVA in linear models; the equivalent is ANODEV in GLMs, which is described somewhere in the edgeR user's guide.

ADD COMMENT
0
Entering edit mode

Aaron, thanks for your response. Follow-up question: using glmFit then, $dispersion gives the tagwise 'within group variance', and $deviance gives the 'between group variance'?

ADD REPLY
0
Entering edit mode

No. Dispersion is analogous to within group variance, but deviance is nothing at all like between group variance. See my separate answer.

ADD REPLY

Login before adding your answer.

Traffic: 832 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