mulink.simulate.hierarchical_mudata

mulink.simulate.hierarchical_mudata(n_mod, *, n_obs=5, n_vertices=2, min_edges=2, extra_edge_probability=0.2, extra_edge_levels=None, transitive_closure=True, varp_key='feature_mapping', random_state=42)

Generate a mudata object with hierarchical feature relationship

Parameters:

• n_mod (int) – Number of modalities (levels) in the object

• n_obs (int (default: 5)) – Number of observations in the object

• n_vertices (int (default: 2)) – Number of vertices in the level with the lowest cardinality

• min_edges (default: 2) – Minimum number of vertices between adjacent levels

• extra_edge_probability (float | None (default: 0.2)) – Probability of adding an additional edge between vertices of adjacent levels. If None, the feature relationship between different levels is represented by a tree. This means that a feature from level n+1 maps to exactly one feature in level n

• extra_edge_levels (list[int] | None (default: None)) – Constrain the addition of extra edges to these levels of the final mudata object (starting at 0). Must not contain the highest level, as these features do not contain additional connections.

• transitive_closure (bool (default: True)) – Whether to return a graph where features that are indirectly linked are connected with an explicit edge. For example, for the feature mapping A –> B –> C, the connection A –> C will also be added

• varp_key (str (default: 'feature_mapping')) – Name of key that contains the feature mapping in .varp attribute of the returned mudata.MuData object

• random_state (int (default: 42)) – Random state for the simulation

Returns:

A mudata.MuData object with explicit feature mapping. The individual modalities/feature-levels are indicated with mod{idx}. The feature mapping is added as adjacency matrix with in the .varp attribute as varp_key. In the matrix, entry (i, j) corresponds to a directed edge from feature i to feature j.