Help with the design matrix for longitudinal study
1
1
Entering edit mode
@sharvari-gujja-6614
Last seen 1 day ago
United States

Hi,

I am interested in using edgeR for DE analysis of longitudinal data. The edgeR manual section 3.3.1 explains how to perform a time series analysis for two conditions at different time points. The design matrix used in the example is:

> design<-model.matrix(~0+Group)


However, if the same subjects were used across time then I wanted to confirm

1. should the design matrix for matched samples in a time series experiment be:
> design <- model.matrix(~subjectID+Group)

2. any change to setting up contrasts for pairwise comparisons against baseline (T0) for each condition and between time points?

Appreciate all the help.

Thanks

edgeR • 94 views
2
Entering edit mode
@gordon-smyth
Last seen 34 minutes ago
WEHI, Melbourne, Australia

Each subject receives all treatments, then you could use

model.matrix(~0+Group+SubjectID)


and the contrasts would be exactly as in Section 3.3.1.

0
Entering edit mode

Thank you so much!