Build a cell2location Model — build_model_cell2location • spacedeconv

Wraps the cell2location Python workflow via reticulate to train a reference model from single-cell data. Returns the trained model object.

Usage

build_model_cell2location(
  single_cell_obj,
  epochs = 250,
  assay_sc = "counts",
  batch_id_col = "sample_id",
  cell_type_col = "celltype_major",
  cell_count_cutoff = 5,
  cell_percentage_cutoff = 0.03,
  nonz_mean_cutoff = 1.12,
  gpu = TRUE
)

Arguments

single_cell_obj: SingleCellExperiment containing reference expression.
epochs: Number of training epochs.
assay_sc: Single-cell assay to use.
batch_id_col: Batch ID column name.
cell_type_col: Column with cell type labels.
cell_count_cutoff: cell2location parameter.
cell_percentage_cutoff: cell2location parameter.
nonz_mean_cutoff: cell2location parameter.
gpu: Logical; whether to train on GPU.