CellScope.ts.generate_tree_structured

CellScope.ts. generate_tree_structured (fea, T, step0 = None, step1 = None, cell_type = None)

The generate_tree_structured function constructs a tree-like representation of cell clustering, identifying hierarchical relationships between clusters across different levels, and generates UMAP-based visualizations for each cluster.

Parameters

• fea (np.ndarray):

  The feature matrix of shape (n_cells, n_genes).

• T (np.ndarray):

  A matrix of clustering labels for different levels of clustering.

• step0 (int, optional, default=None):

  The starting level for the tree-structured visualization. If None, it is automatically determined.

• step1 (int, optional, default=None):

  The ending level for the tree-structured visualization. If None, it is determined based on the highest ARI score.

• cell_type (np.ndarray, optional, default=None):

  A 1D array of cell type labels. Required if step1 is not provided.

Return

• Y_initial (np.ndarray):

  The initial 2D UMAP embedding of all cells.

• label_step0 (np.ndarray):

  Clustering labels at the starting level (step0).

• Y_1 (list of np.ndarray):

  A list of UMAP embeddings for clusters at the first level.

• Title_1 (list):

  A list of titles for the first-level clusters.

• Y_all (list of np.ndarray):

  A list of UMAP embeddings for clusters at all levels.

• Title_all (list):

  A list of titles for clusters at all levels.

• index_1 (list):

  A list of indices corresponding to cells in each cluster at the first level.

• index_all (list):

  A list of indices corresponding to cells in each cluster at all levels.

• step0 (int):

  The final starting level for the tree-structured visualization.

• step1 (int):

  The final ending level for the tree-structured visualization.