Dear all,
I am using q2-SCNIC to analyze correlation networks on my data, and I have some questions related with it.
I am comparing the microbiota of healthy (N=7) and diseased (N=6) animals at three time points: days 0, 30 and 60 (disease was around day 17). At the comparison between healthy and disease samples, I found no statistically significant results on alpha and beta diversity, as well as in any differential abundance approach. Maybe, the low number of samples has something to do with this...
However, when I checked the correlation network to seek for possible hidden differences, the differences in the complexity of the networks are very different and difficult to interpret. All newtorks were computed using a genus-level-collapsed table, sparcc method (for compositional data), and same filtering (edge r > 0.85, node degree > 2). To illustrate differences I provide nodes and edges of each network, but I can provide other values if needed.
| DAY/HEALTH STATUS | NODES | EDGES |
|---|---|---|
| 0 HEALTHY | 47 | 78 |
| 0 DISEASE | 59 | 134 |
| 30 H | 104 | 262 |
| 30 D | 104 | 477 |
| 60 H | 67 | 142 |
| 60 D | 122 | 819 |
(Differences inside Healthy samples (day 30 vs 60), I'll try to solve them later)
My main questions are:
1: Does it have sense to look for differences in the correlation network after getting no statistically significant results in alpha, beta, etc.?
2: what could be the explanation to having such differences between Healthy and Disease networks after having no differences at equal time points?
Thanks a lot in advance!!