lynx.dataset.XeniumDataset

class lynx.dataset.XeniumDataset(*args, **kwargs)

Bases: Dataset

Load Xenium ST graph w/ auxiliary DESI modality

Parameters:

• adatas (Union[sc.AnnData, List[sc.AnnData]]) – List of spatial data

• k (int) – Resolution for k-NN graph

• n_subgraphs (int) – Number of graph partition batches per data

• r (float)

• is_grid (bool)

• verbose (bool)

__init__(adatas, k=8, r=50, n_subgraphs=8, is_grid=False, verbose=True, **kwargs)

Parameters:

• adatas (scanpy.AnnData | List[scanpy.AnnData])

• k (int)

• r (float)

• n_subgraphs (int)

• is_grid (bool)

• verbose (bool)

Methods

__init__(adatas[, k, r, n_subgraphs, ...])
construct_graph(neighbor_nodes, distances[, ...])	Compute undirected graph edges & attributes
get(idx)
get_neighbors(ref_coords, query_coords[, ...])	Retrieve k-nearest-neighbor (or radius-bounded neighbors) of query_coords to ref_coords using a KDTree
len()
load_graphs()

construct_graph(neighbor_nodes, distances, cluster_labels=None)

Compute undirected graph edges & attributes

Parameters:

• neighbor_nodes (ndarray | torch.tensor | list)

• distances (ndarray | torch.tensor | list)

• cluster_labels (ndarray | None)

static get_neighbors(ref_coords, query_coords, is_grid=False, k=8, r=50)

Retrieve k-nearest-neighbor (or radius-bounded neighbors) of query_coords to ref_coords using a KDTree

Parameters:

• ref_coords (ndarray | torch.tensor | list)

• query_coords (ndarray | torch.tensor | list)

• is_grid (bool)

• k (float)

• r (float)