Bayesian methods have been enormously helpful in analyzing temporally regulated genes from microarray datasets. See "An improved empirical bayes approach to estimating differential gene expression in microarray time-course data: BETR (Bayesian Estimation of Temporal Regulation)" http://www.biomedcentral.com/1471-2105/10/409 for example (additional examples abound in the literature).

In principle, I think you could apply the R package developed in the above paper to your RNA-Seq datasets if you convert your data to a single metric per gene locus.

I agree it is critical that you avoid simply looking at pairwise comparisons if the important biology is the overall temporal trend.

Thanks a lot, I'll try to apply the approach you suggested. I also found that cuffdiff has a time-series flag; do you think it may be worth a try?

I haven't used cuffdiff, but there is some discussion about it at this thread:

https://www.researchgate.net/post/What_is_your_favorite_DEG_test_for_RNA-seq_data