I am new in field of RNA-Seq and now trying to call DEGs from 2 plant genotypes (A -heat resistant and B-heat unresistant). I have 2 treatments (control vs treatment with 3 replicates for each, except one control I have only 2 replicates) and 2 timepoints (6 days and 12 days). Basically I would like to:
Compare within and across timepoints for each genotype, e.g. A_treatment_day6 - A_control_day6 and (A_treatment_day12 - A_control_day12) - (A_treatment_day6 - A_control_day6)
Compare within and across timepoints across genotypes (e.g. (A_treatment_day6 - A_control_day6I) - (B_treatment_day6 - B_control_day6)).
I am very confused whether it is better to build one matrix with genotypes, timepoints and treatment (~0+genotype+timepoint+treatment) and call DEGs from it, or it is better to define each treatment combination as a group (~0 + group; as in the section 3.3.1 of edgeR manual). Since the transcriptional responses can be quite different at 6 and 12 days, another thought I have is that perhaps it is better to work on each timepoint separately and treat each treatment combination as a group? But how can I avoid creating the (~0+genotype+timepoint+treatment) matrix for (2)?
May I share your thoughts what would be the best approach for this analysis? I read through edgeR manual a few times, but with my limited statistical knowledge I am not sure I understand it by heart. I greatly appreciate all suggestions and guidance. Thank you very much and I am looking forward to hearing from you.