cellink.ml.MILDataset

class cellink.ml.MILDataset(dd, donor_layer=None, donor_labels_key=None, donor_batch_key=None, donor_cat_covs_key=None, donor_cont_covs_key=None, donor_indices_key=None, donor_id_key=None, celltype_key=None, cell_layer=None, cell_labels_key=None, cell_batch_key=None, cell_cat_covs_key=None, cell_cont_covs_key=None, cell_indices_key=None, cell_donor_key=None, split_donors=None, split_indices=None)

PyTorch Dataset for Multiple Instance Learning (MIL) using donor- and cell-level data from AnnData or MuData objects.

This dataset supports both pre-split donor indices and random splitting. Each item returned by the dataset contains the donor-level and corresponding cell-level features and labels, along with optional batch information and covariates.

Methods table

Methods