Dimension reduction techniques (t-SNE, UMAP) in Haill

Hello everyone. i would like to introduce a new topic on dimension reduction techniques. Particularly i would like to introduce new statistical methods to Hill. As far as i understand currently only one dimension reduction technique (PCA) is introduced in Hail. I wonder is there another dimension reduction techniques like t-SNE or UMAP? If it is interested to the audience we could discuss an opportunity to introduce more statistical techniques to the Hill

Hey @Alexey ! We are indeed interested in implementing these techniques atop the Hail platform. We’re currently investigating the technical feasibility of this. The primary challenge is adapting these highly iterative linear algebraic techniques to Hail’s computational model. @patrick-schultz may have more to share here.