Question: gene expression variance equality
gravatar for galozs
3.0 years ago by
galozs20 wrote:


I am analyzing RNA seq data and trying to identify differentially expressed genes between to populations using ttest (my data is in triplicates)

I was wondering weather I need to check for the variance assumptions before applying ttest.

if so- what test will be recommended to use on such data?

I have tried Bartlett's, Levene's (with absolute values), Levene's (with quadratic values), Brown-Forsythe and two-sample F-test

and got different though close results. I would live to hear your thoughts! Tnx!

ADD COMMENTlink modified 3.0 years ago by e.rempel900 • written 3.0 years ago by galozs20
gravatar for e.rempel
3.0 years ago by
Germany, Heidelberg, COS
e.rempel900 wrote:


citation from this thread (answer from Gordon Smyth - author of limma):

Unequal variances are not usually a major concern in microarray data. The two-sample t-test is known to be highly robust against unequal variances, and the limma moderated t-test inherits this property.

Thus, I would recommend that you run limma (or DESeq2 or EdgeR) for your data.

ADD COMMENTlink written 3.0 years ago by e.rempel900

I am looking at exactly 2 conditions and prefer working with ttest and my own codes.


ADD REPLYlink written 3.0 years ago by galozs20

May I ask why? limma and other tools inherit the "good" properties of t-Test and have some additional advantages. Personally, I would use t-Test in case of (real) many replicates and a few genes.

ADD REPLYlink written 3.0 years ago by e.rempel900

honestly, I used cuffdiff before and realized that using my own code and analysis gives more reliable results. I didn't try the other tools you mentioned, yet. maybe I'll give it a try in the future.

ADD REPLYlink written 3.0 years ago by galozs20
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1710 users visited in the last hour