Hi @gregcaporaso,
Yes, the easiest way to use unifrac is
qiime diversity core-metrics-phylogenetic.
I faced a roadblock in this as I needed to generate rep-seqs for each body site but I used the grouped table as input to do so. This subsequently led to problems down the road while running core-metrics-phylogenetic as it was not recognizing the subject column so I had to replace all the sample IDs with the respective subjects to run core-metrics-phylogenetic (Hopefully this was the right thing to do).
Here are the results, however the p values are still greater than 0.05. Does a high p value mean there is no correlation between the two sites after factoring in individual subjects? Based on my (limited) understanding, the null hypothesis in Mantel Test is that distance matrix 1 is independent of distance matrix 2, therefore in this case, null hypothesis is not rejected and the distance matrices are independent. (Side Question: What if p value was less than 0.05? Does that mean there is some dependence (correlation) between distance matrices?)
Unweighted Unifrac:
| Mantel test results | |
|---|---|
| Method | Spearman |
| Sample size | 6 |
| Permutations | 999 |
| Alternative hypothesis | two-sided |
| Spearman rho | 0.425 |
| p-value | 0.097 |
Weighted Unifrac:
| Mantel test results | |
|---|---|
| Method | Spearman |
| Sample size | 6 |
| Permutations | 999 |
| Alternative hypothesis | two-sided |
| Spearman rho | 0.007143 |
| p-value | 0.983 |
Bray Curtis:
| Mantel Test Results | |
|---|---|
| Method | Spearman |
| Sample size | 7 |
| Permutations | 999 |
| Alternative hypothesis | two-sided |
| Spearman rho | 0.01236 |
| p-value | 0.961 |
Thank you again!