scFates.tl.cluster¶
- scFates.tl.cluster(adata, layer='fitted', n_neighbors=20, n_pcs=50, metric='cosine', resolution=1, device='cpu', copy=False)¶
Cluster features. Uses scanpy backend when using cpu, and cuml when using gpu. Dataset is transposed, PCA is calulcated and a nearest neighbor graph is generated from PC space. Leiden algorithm is used for community detection.
- Parameters
- adata :
AnnData
Annotated data matrix.
- layer
Layer of feature to calculate clusters, by default fitted features
- n_neighbors :
int
(default:20
) Number of neighbors.
- n_pcs :
int
(default:50
) Number of PC to keep for PCA.
- metric :
str
(default:'cosine'
) distance metric to use for clustering.
- resolution :
float
(default:1
) Resolution parameter for leiden algorithm.
- device :
str
(default:'cpu'
) run the analysis on ‘cpu’ with phenograph, or on ‘gpu’ with grapheno.
- copy :
bool
(default:False
) Return a copy instead of writing to adata.
- adata :
- Returns
adata – if copy=True it returns subsetted or else subset (keeping only significant features) and add fields to adata: .var[‘cluters’]
module assignments for features.
- Return type