Dear all,

I am a newbie to RNA-seq analysis related to samples collected at different time points.

I have 6 goats and collected RNA-seq samples for each animal at 4 different time points within one year. The samples are collected at different time points where the environmental temperature various (for e.g. Autumn, winter, spring and summer). I would like to identify differentially expressed genes associated with seasonal changes.

I am highly apricate if someone give me a tip how to analyze my data. In addition, please share me any good tutorial related to my data.

Br, Get.

Typically such a setup assumes that gene expression is a smooth trend across the seasons rather than making discrete jumps. A common strategy is to use cosinor regression or similar approaches. Check on tools such as MetaCycle, limorhyde, discoRhythm, JTKcycle and see this workflow which covers these sorts of designs. All mentioned tools are typically used for circadian analysis (gene expression rhythmicity over 24h) but essentially it's the same here, just modelling over 12 months. If you any reason none of the tools work for you or you do not detect rhythmicity (for whatever reason) you can still go back doing either pairwise comparisons between the seasons or use the LRT of DESeq2 to find any difference associated with the seasons. I would go for the rhythmicity approach and mentioned tools first, that is common in the field.

For guidance, see some landmark studies (human, not goat but concept is the same). These two papers are typically cited, despite that they have quite a large n and more than just four timepoints. The rhythmicity estimates might be a bit bulky with just four timepoints.

• https://www.nature.com/articles/ncomms8000
• https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7529769/

What you can of course also do is to run the pairwise comparisons and see in a heatmap whether you see clusters of genes that somewhat resemble a trended seasonal expression. That is less "strict" than cosinor regression but might get some genes in case the "standard" method do not yield anything.

Note that this is all my opinion and experience, I am not the DESeq2 author and hence this is no official advise.