scmagnify.tools.build_metacells_SEACells

scmagnify.tools.build_metacells_SEACells#

scmagnify.tools.build_metacells_SEACells(mdata, rna_key='RNA', atac_key='ATAC', rna_dr_key='X_pca', atac_dr_key='X_svd', use_raw=False, rna_layer='counts', atac_layer='counts', n_metacells=None, min_iter=10, max_iter=50, use_gpu=False, groupby='celltype', mask_key='cell_state_masks', embed_key='X_umap', t_key='palantir_pseudotime')#

Build metacells using SEACells for RNA and ATAC modalities in a MuData object.

Parameters:
  • mdata (MuData) – MuData object containing RNA and ATAC modalities.

  • rna_key (str (default: 'RNA')) – Key for RNA modality in mdata.

  • atac_key (str (default: 'ATAC')) – Key for ATAC modality in mdata.

  • rna_dr_key (str (default: 'X_pca')) – Dimension-reduction key for RNA (e.g., ‘X_pca’).

  • atac_dr_key (str (default: 'X_svd')) – Dimension-reduction key for ATAC (e.g., ‘X_svd’).

  • use_raw (bool (default: False)) – Whether to use .raw.

  • rna_layer (str | None (default: 'counts')) – Layer name to use for RNA counts; if None, uses X.

  • atac_layer (str | None (default: 'counts')) – Layer name to use for ATAC counts; if None, uses X.

  • n_metacells (Optional[int] (default: None)) – Number of metacells; if None, determined from cell count.

  • min_iter (int (default: 10)) – Minimum SEACells iterations.

  • max_iter (int (default: 50)) – Maximum SEACells iterations.

  • use_gpu (bool (default: False)) – Whether to use GPU.

  • groupby (str | None (default: 'celltype')) – obs column for cell-type annotation; majority label per metacell is recorded if provided.

  • mask_key (str | None (default: 'cell_state_masks')) – obsm key for cell-state masks to propagate to metacells if present.

  • embed_key (str | None (default: 'X_umap')) – obsm embedding key used for visualization.

  • t_key (str | None (default: 'palantir_pseudotime')) – obs pseudotime key.

Return type:

MuData

Returns:

MuData MuData object containing metacells.