Model¶
The heterogeneous attention VGAE and its configuration / dataset helpers.
Learning latent manifold w/ Conditional VGAE on hetero-graph Generative path: DESI (u) -> Latent (z) -> Xenium (x) |
|
Load paired multi-modal ST data w/ hybrid resolutions into a hetero-graph |
|
Load Xenium ST graph w/ auxiliary DESI modality |
|