Are polynomial contrasts for a time factor possible in DESeq2?
1
1
Entering edit mode
dairey ▴ 10
@794de4b7
Last seen 12 months ago
United States

Are polynomial contrasts for a time factor possible in DESeq2?

If we have a design with multiple time points being measured for independent samples, and we want to ask about linear or quadratic trends, is it possible to use polynomial contrasts? For example the contrasts for 4 time points would be:

Linear: -3, −1, 1, 3

Quadratic: 1, −1, −1, 1

Cubic: −1, 3, −3, 1

We have already compared adjacent time points using the contrast command in DESeq2. But someone asked about testing the trend over these time points. Using the linear polynomial contrast above is one solution, although admittendly I've applied these before in a linear regression setting and not a negative binomial regression. I suppose with 4 time points we could also treat time as continuous.

Thank you for any comments.

Dave A.

DESeq2 • 360 views
ADD COMMENT
1
Entering edit mode
ATpoint ★ 3.9k
@atpoint-13662
Last seen 13 minutes ago
Germany

As you can use a model.matrix() as input for the design you can use a great variety of models. A great read is this work from Ritchie/Smyth labs which also covers quadratic/polynomial models: labhttps://bioconductor.org/packages/release/workflows/vignettes/RNAseq123/inst/doc/designmatrices.html

A reference for a published package using DESeq2 for a model as in the above link (circadian/oscillation model) is compareRhythms.

ADD COMMENT

Login before adding your answer.

Traffic: 635 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