Hi, I'm new to bioinformatics. I have a weighted UniFrac metric generated from some bacterial samples and I want to test the effect of two factors A and B (both binary) on the composition using adonis2 in R. When I only include factor A, the result is not significant. But if I include the interaction of A and B, the interaction of A*B is not significant, but the individual effect of A and B are both significant. I'm wondering why the effect of A becomes significant, cause from the PCoA plot, the four groups (i.e., A1*B1, A1*B2, A2*B1, A2*B2) somehow overlapped.
Hey, there's an important point to remember about ordination techniques like PCoA and NMDS. While they're great for visualizing complex data in 2D or 3D, they're just a simplified snapshot.
Look at the contribution % of each axes or the stress values (NMDS). It can some time help to try different ordination tech.
Another point to consider is the number of samples per condition A and B, maybe with more sample the cleavage will be more visible.
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics