celcomen.datareaders.datareader.get_dataset_loaders

celcomen.datareaders.datareader.get_dataset_loaders(h5ad_path: str, sample_id_name: str, n_neighbors: int, distance: float, device: str, verbose: bool)

Prepares and returns PyTorch Geometric DataLoader from a single-cell spatial transcriptomics dataset.

The function reads a single-cell AnnData object from an H5AD file, normalises the data, and generates graph data where nodes correspond to cells, and edges are created based on spatial proximity using a k-nearest neighbours graph. The data is then loaded into a PyTorch Geometric DataLoader.

Parameters

h5ad_pathstr: Path to the H5AD file containing the raw counts of the single-cell spatial transcriptomics data.
sample_id_namestr: Name of the sample ID column in adata.obs to separate the dataset into different samples.
n_neighborsint: Number of neighbours to use for constructing the k-nearest neighbours graph for spatial information.
distancefloat: Distance of neighbours to use for constructing the k-nearest neighbours graph for spatial information.
verbosebool: If True, prints detailed information about the DataLoader during the loading process.

Returns

DataLoader: A PyTorch Geometric DataLoader containing the processed graph data, with each graph representing a sample of cells in the dataset.

Notes

The spatial positions of the cells are used to create a k-nearest neighbours graph, with edges connecting cells that are spatially close to each other.
The input features for the graph (x) are normalised before constructing the graph.
adata.obsm[“spatial”] is used to extract the spatial coordinates of the cells.
The graph data is validated using PyTorch Geometric’s built-in validation method.

Examples

>>> loader = get_dataset_loaders('data.h5ad', 'sample_id', n_neighbors=6, verbose=True)
Step 1
=====
Number of graphs in the current batch: 1
Data(x=[100, 33500], edge_index=[2, 500], pos=[100, 2], y=[1])

Raises

ValueError: If there are issues with the input data during validation, e.g., if the graph is not well-formed.