Entering edit mode
In a differential abundance modelling using DESeq2, given a single ASV (out of about 1000) with the following abundance values:
| Seq 3 | ||
|---|---|---|
| 1 | A1-2 | 26250 |
| 2 | A1-3 | 45729 |
| 3 | A1-4 | 56033 |
| 4 | A1-5 | 62823 |
| 5 | A1-6 | 53235 |
| 6 | A1-7 | 43591 |
| 7 | A1-8 | 1592 |
| 8 | A4-2 | 111046 |
| 9 | A4-3 | 67847 |
| 10 | A4-4 | 101865 |
| 11 | A4-5 | 5869 |
| 12 | A4-6 | 41578 |
| 13 | A4-8 | 66016 |
With A1 being the treatment and A4 the reference (the first level on the factor assigned to colData). The mean A1 is 41,322 and mean A4 is 65,703. How could it be then that the model shows a positive LFE of 1.96 and a significance of P = 6.8e-04?
What is surprising to you? Please see DESeq2 vignette on what is the reference level considered in the differential analysis test : "if you never tell the DESeq2 functions which level you want to compare against (e.g. which level represents the control group), the comparisons will be based on the alphabetical order of the levels". Then A1 is probably the reference in your analysis but you do not show the code you performed, you must improve your post to clarify your question.
Thanks for your comment. Indeed, I set A4 to be the reference manually using
fct_relevel(., ref_level)and also use thecontrastargument when callingresults. So I'm quite convinced that this is not the issue. I'm guessing it probably has to do with the overall average in the sample across all the ASVs