I had a “methodology” question to ask you. Let’s say I have three samples: one uninfected (control), one infected at 24 hours, and a third at 48 hours. I want to identify the differentially expressed genes between the 24-hour and 48-hour conditions. What’s the best experimental design?

- Run an edgeR analysis with a `sample_list` containing only two conditions—for example, 24h vs. T—and another analysis (with a different `sample_list`) comparing 48h vs. T; basically, perform two separate edgeR analyses, or

- Run a single edgeR analysis with a single `sample_list` containing all three conditions: 24h, 48h, and T?

Naively, I thought the differential expression values would be the same—for example, for the 24h condition in both tests—but in fact, the presence of the “48h” sample in the second test alters the differential expression values for the 24h condition in that test.

So I was thinking of sticking with Test 1 using two EdgeR runs in parallel (24h vs. T and 48h vs. T), then comparing the log2FC values from the two tests—what do you think?