I think the point of using t-SNE was to suggest that if the structure were legitimate that t-SNE would find at least some of it. This is both an interesting/fu

I think the point of using t-SNE was to suggest that if the structure were legitimate that t-SNE would find at least some of it. This is both an interesting/fun visualization exercise and a cautionary story. Apparently UMAP has a tendency to render blobs as rings or loops!