methods in aligned umap

[waitati]

I have two data sets, a source (S) and a target (T). I want to use umap align to create low dimensional maps that
transform S and T to low dimensional projections that are aligned. I now want to take a new data point
in S and map it to 's' (its low dimensional representation) and then from 's' to 't', using essentially
an identity map produce by aligned umap, and then back up to a target in T.

For that, i think I need to have a method in umap align that corresponds to the 'transform' methods
in umap, and the inverse estimator in umap. I do not think either of these methods exist in umap align.
Is that correct?

Is it possible to 'pull out' the transform used for S, and the transform used for T, from the umap aligned
object? If so, I can then apply the map for S to the input data, and then try to create an inverse
transform for T to go from t back up to T.

Hope this is kinda of clear? Basically, I just want to use umap aligned to train a model that predicts
unseen source data mapped to a target prediction given labeled pairs of the the source and target
data for training.

Thank you
nick