omicverse.pp.mde

omicverse.pp.mde(adata, embedding_dim=2, n_neighbors=15, basis='X_mde', n_pcs=None, use_rep=None, knn=True, transformer=None, metric='euclidean', verbose=False, key_added=None, random_state=0, repulsive_fraction=0.7, constraint=None)[source]

Run MDE (Minimum Distortion Embedding) from a latent representation.

Parameters:

  • adata (anndata.AnnData) – AnnData object containing a latent representation in adata.obsm.

  • embedding_dim (int, default=2) – Number of dimensions in the output embedding.

  • n_neighbors (int, default=15) – Number of neighbors used for graph construction in MDE.

  • basis (str, default="X_mde") – Key in adata.obsm where the result is stored.

  • n_pcs (int, optional) – Number of principal components used from use_rep.

  • use_rep (str, optional) – Input representation key in adata.obsm. Defaults to 'X_pca'.

  • knn (bool, default=True) – Whether to use k-nearest-neighbor graph initialization.

  • transformer (object, optional) – Optional neighbor transformer object.

  • metric (str, default="euclidean") – Distance metric for neighborhood search.

  • verbose (bool, default=False) – Whether to print detailed optimization logs.

  • key_added (str, optional) – Optional key alias for metadata outputs.

  • random_state (int, default=0) – Random seed for reproducibility.

  • repulsive_fraction (float, default=0.7) – Repulsion weight controlling global separation.

  • constraint (object, optional) – PyMDE constraint object. Uses pymde.Standardized() when None.

Returns:

Stores the low-dimensional embedding in adata.obsm[basis].

Return type:

None