I created a PCoA on weighted UniFrac distance and felt that it would be useful to find the loadings of the PCoA axes. I might want to find which features are taking higher weights along a specific axis. Does anyone know how to check the loadings?
Unfortunately with principal coordinates analysis there’s no way to access loading information (as in principal component analysis). Recall that PCoA is computed based on a distance matrix, and distance functions preserve no information about the features that make up the distances.
Although this isn’t equivalent to PCoA loadings, you can compute PCoA biplots (currently not available in QIIME but will hopefully make its way to the next release ©). The biplots would tell you what features are correlated with what axes of variation and what samples.
Another alternative would be to use something like EMDUniFrac. Based on UniFrac distances, it can tell you about discriminant branches in your tree (paper).
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.