Closed TuomasBorman closed 11 months ago
Sounds good to me at least and we can always refine later if new needs will arise. The default is something worth considering - which one that should be.
So what I understood https://rdrr.io/rforge/vegan/man/cca.object.html
Ordination scores:
u --> correlation between sample metadata variables v ---> correlation between features wa --> u weighted with v
In this case, I think we should allow these all options, but wa should be the default
wa can be default; this is "scores" and "loadings" from the model specification, rather than correlations in traditional sense but yes.
New parameter scores
which can get values "wa"
or "u"
.
CCA includes linear combinations ("u") values and weighted average ("wa") values. Currently, linear combinations are returned in runCCA, although, cca object is as an attribute, so all the data can be accessible from the cca object.
However, plotReducedDim for instance plots the matrix in reducedDim meaning that it plots the linear combination values even though weighted averages might be more interesting.
I added an option to choose the desired values (weighted averages as default).
Comments?