Dear all,
I have an experiment addressing gene expression differences during different stages of developmental time line (stage Early, Mid and Late) across two consecutive seasons (Fall and Spring). Each sample of the timeline was taken for the same individual, for a total of 4 individuals for all seasons and stages, and so our sampling design looks like so:
Individual Season Stage
1 Fall Early
1 Fall Mid
1 Fall Late
2 Fall Early
2 Fall Mid
2 Fall Late
3 Fall Early
3 Fall Mid
3 Fall Late
4 Fall Early
4 Fall Mid
4 Fall Late
1 Spring Early
1 Spring Mid
1 Spring Late
2 Spring Early
2 Spring Mid
2 Spring Late
3 Spring Early
3 Spring Mid
3 Spring Late
4 Spring Early
4 Spring Mid
4 Spring Late
I am interested in genes that show differential expression across stages within each season, since we’re interested in genes that are, say, important for the induction of the developmental timeline (up regulated in early) in general. But also, I am interested in genes that show differential expression between stages across seasons. For instance, I’m also interested in genes that show differences in expression between “early” stages between Fall and Spring, to see what genes are activated (or more expressed) in one season and not the other.
I have looked into the tutorial for time-course experiments of DESeq2, and searched some of the forums (http://seqanswers.com/forums/showthread.php?t=50007 and https://bioconductor.org/packages/release/workflows/vignettes/rnaseqGene/inst/doc/rnaseqGene.html#time-course-experiments), but I’m not sure how to the correlated individual samples should be treated, or if my models are removing effects that I am interested in as well.
For instance, I’ve constructed two models:
M1(full) ~stage + season + stage:season
M2(reduced) ~stage + season
But I wonder if this will remove the within season differences between stages, which I am interested in as well, and how would these models account for the non-independence of samples of the same individual.
Thank you very much in advance.
Cheers.
Dear Michael,
Yes, those samples belong to the same individual.
In fact, this was a simplified example of our dataset. We have some additional individuals that we would like to add to this dataset, which were not sampled for all stages and/or were sampled only for one season. Nevertheless, we could subsample the dataset to have a more balance sampling (which is more in line with the example we gave).
This is the full dataset and design looks like so:
Sample_ID Season Stage
1 Spring Early
1 Spring Mid
1 Fall Early
1 Fall Late
2 Spring Early
2 Spring Mid
2 Spring Late
2 Fall Mid
2 Fall Late
3 Spring Early
3 Spring Mid
3 Spring Late
3 Fall Mid
3 Fall Late
4 Spring Early
4 Spring Mid
4 Spring Late
5 Spring Mid
5 Spring Late
5 Fall Early
5 Fall Mid
5 Fall Late
6 Fall Early
6 Fall Mid
7 Fall Early
7 Fall Mid
7 Fall Late
Thank you for your prompt response.