LRT, more than two conditions and pairwise comparisons
1
0
Entering edit mode
boczniak767 ▴ 740
@maciej-jonczyk-3945
Last seen 8 days ago
Poland

Hi,

I've read the instruction for performing likelihood ratio test. The example is clear but in the test dataset there are only two conditions (wildtype and mutant).

I have dataset with data for four maize lines two tissues and 12 time-points. I wonder if it is possible to make pairwise comparisons using full dataset.

Or should I

  1. Compute size-factors for full DESeqDataSet and use formula ~line + time + line*time (comparisons across tissues are not of interest)
  2. Subset the full dataset to contain only one tissue and two lines and compare them.
  3. Repeat step 2. for each pairwise comparison within each tissue
  4. Do FDR correction for all p-values ?
DESeq2 • 265 views
ADD COMMENT
0
Entering edit mode
@james-w-macdonald-5106
Last seen 7 hours ago
United States

Unless you think the different lines or tissues will have very different within-group variances, you should just analyze together and make comparisons you care about using Wald statistics. The LRT is meant as an omnibus test to determine if a larger model fits better than a reduced model, and is not really applicable to what you are trying to do.

0
Entering edit mode

Thank you for response, apparently I still don't know much about statistics. In fact I want to campare time-profiles and the LRT has been proposed for this in the tutorial. If I use Wald test I'd have to compare time-points individually.

ADD REPLY
1
Entering edit mode

In that case, probably easiest to subset to the groups being compared.

ADD REPLY

Login before adding your answer.

Traffic: 372 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