scanpy.tl.draw_graph

scanpy.tl.draw_graph(adata, layout='fa', *, init_pos=None, root=None, rng=None, n_jobs=None, adjacency=None, key_added=None (sc.settings.preset='scanpy-v1' – changes in 2.0), neighbors_key=None, obsp=None, copy=False, key_added_ext=None, **kwds)

Force-directed graph drawing [Chippada, 2018, Islam et al., 2011, Jacomy et al., 2014].

An alternative to tSNE that often preserves the topology of the data better. This requires running neighbors(), first.

The default layout (‘fa’, ForceAtlas2, Jacomy et al. [2014]) uses the package fa2-modified [Chippada, 2018], which can be installed via pip install fa2-modified.

Force-directed graph drawing describes a class of long-established algorithms for visualizing graphs. It was suggested for visualizing single-cell data by Islam et al. [2011]. Many other layouts as implemented in igraph [Csárdi and Nepusz, 2006] are available. Similar approaches have been used by Zunder et al. [2015] or Weinreb et al. [2017].

Array type support

Array type	supported	… experimentally in dask Array
numpy.ndarray	✅	❌
scipy.sparse.{csr,csc}_{array,matrix}	✅	❌

Parameters:

adata AnnData

Annotated data matrix.

layout Literal['fr', 'drl', 'kk', 'grid_fr', 'lgl', 'rt', 'rt_circular', 'fa'] (default: 'fa')

‘fa’ (ForceAtlas2) or any valid igraph layout. Of particular interest are ‘fr’ (Fruchterman Reingold), ‘grid_fr’ (Grid Fruchterman Reingold, faster than ‘fr’), ‘kk’ (Kamadi Kawai’, slower than ‘fr’), ‘lgl’ (Large Graph, very fast), ‘drl’ (Distributed Recursive Layout, pretty fast) and ‘rt’ (Reingold Tilford tree layout).

root int | None (default: None)

Root for tree layouts.

rng int | integer | Sequence[int] | SeedSequence | Generator | BitGenerator | None (default: None)

Random number generation to control stochasticity.

If a type:SeedLike value, it’s used to seed a new random number generator; If a numpy.random.Generator, rng’s state will be directly advanced; If None, a non-reproducible random number generator is used. See numpy.random.default_rng() for more details.

The default value matches legacy scanpy behavior and will change to None in scanpy 2.0.

Applies to layouts with random initialization like 'fr'.

adjacency csr_array | csc_array | csr_matrix | csc_matrix | None (default: None)

Sparse adjacency matrix of the graph, defaults to neighbors connectivities.

key_added str | None | Default (default: None (sc.settings.preset='scanpy-v1' – changes in 2.0))

Template for the key. If None, uses f'X_draw_graph_{layout}' for obsm.

proceed

Continue computation, starting off with f'X_draw_graph_{layout}'.

init_pos str | bool | None (default: None)

'paga'/True, None/False, or any valid 2d-.obsm key. Use precomputed coordinates for initialization. If False/None (the default), initialize randomly.

neighbors_key str | None (default: None)

If not specified, draw_graph looks at .obsp['connectivities'] for connectivities (default storage place for pp.neighbors). If specified, draw_graph looks at .obsp[.uns[neighbors_key]['connectivities_key']] for connectivities.

obsp str | None (default: None)

Use .obsp[obsp] as adjacency. You can’t specify both obsp and neighbors_key at the same time.

copy bool (default: False)

Return a copy instead of writing to adata.

**kwds

Parameters of chosen igraph layout. See e.g. layout_fruchterman_reingold() [Fruchterman and Reingold, 1991]. One of the most important ones is maxiter.

Return type:

AnnData | None

Returns:

Returns None if copy=False, else returns an AnnData object. Sets the following fields:

adata.obsm[(f'X_draw_graph_{layout}' | key_added).format(layout=layout)]numpy.ndarray (dtype float)

Coordinates of graph layout. E.g. for layout='fa' (the default), the field is called 'X_draw_graph_fa'. key_added_ext overwrites layout.

adata.uns['draw_graph']: dict

draw_graph parameters.