nGraph

nGraph

nGraph(
    data,
    layer="raw",
    standardScale=False,
    runPCA=False,
    kNeighbors=15,
    nPcs=20,
    sdataTable=None,
)

Build a k-nearest-neighbour graph from the expression matrix.

Returns a bare igraph.Graph rather than modifying the object, so it is a building block: sp.tl.cluster with method='leiden' builds its own graph internally, and this is here for when you want to run igraph's own algorithms — betweenness, communities, shortest paths — over the same structure.

Parameters:

Name	Type	Description	Default
data	(AnnData | SpatialData | str, required)	The cell table. An AnnData, a path to an .h5ad file, a SpatialData object, or a path to a .zarr SpatialData store.	required
layer	str	Which matrix to build the graph from. 'raw' uses .raw.X, None uses .X, any other string names a key in .layers.	'raw'
standardScale	bool	Z-score each marker before computing neighbours, so markers with large dynamic range do not dominate the distances.	False
runPCA	bool	Reduce to nPcs principal components first. Worth doing above a few dozen markers.	False
kNeighbors	int	Neighbours per cell.	15
nPcs	int	Components to keep when runPCA=True. Clipped to what the data allows.	20
sdataTable	str	Which SpatialData table to read.	None

Returns:

Name	Type	Description
graph	Graph	An undirected graph with one vertex per cell, in obs order.

Raises:

Type	Description
ValueError	When layer='raw' and the object has no .raw.

Example

graph = sp.pp.nGraph(adata, layer="raw", standardScale=True, runPCA=True)
graph.vcount(), graph.ecount()

# Any igraph algorithm now applies.
communities = graph.community_multilevel()
adata.obs["igraph_community"] = [str(c) for c in communities.membership]