edgeR Time course analysis, choice on degree of freedom
1
1
Entering edit mode
Ruixuan ▴ 10
@ruixuan-23626
Last seen 2.2 years ago
Kyoto University

Hi everyone. I have a question on using edgeR to analyze time series data. How to choose the degree of freedom when making design matrix? I basically follow the manual section 4.8 Time course RNA-seq experiments of Drosophila melanogaster to do the analysis;

My sample is here, a 6 time points experiment without replicates. I want to compare specific time effect

Hours <- c(0,1,2,4,8,16)
Time <- paste0(Hours,"hrs")
y <- DGEList(counts=x, group = Time)


To create a design matrix, both poly() and ns() need a degree of freedom. I'm new to this field, what value should I use here? I also find that in the later part 4.8.8 Time course trend analysis, it says that "We test for a trend by conducting F-tests on 3 df for each gene:", does this df in F-test relate to the design matrix one?

Thank you!

edger • 289 views
3
Entering edit mode
@gordon-smyth
Last seen 10 hours ago
WEHI, Melbourne, Australia

Since you're following the User's Guide, just use the same df as used in the case study. The same choice will work for most datasets.

Yes, the F-test is on the same df as the spline. The F-test is testing whether there is a non-zero spline trend.

0
Entering edit mode