omicverse.bulk2single.BulkTrajBlend

omicverse.bulk2single.BulkTrajBlend#

class omicverse.bulk2single.BulkTrajBlend(bulk_seq, single_seq, celltype_key, bulk_group=None, max_single_cells=5000, top_marker_num=500, ratio_num=1, gpu=0)[source]#

Integrate bulk and single-cell information to infer transitional cell-state trajectories.

Parameters:

  • bulk_seq (pd.DataFrame) – Bulk expression matrix with genes in rows and samples in columns.

  • single_seq (anndata.AnnData) – Reference single-cell dataset used to define cell identities.

  • celltype_key (str) – Column name in single_seq.obs containing cell-type labels.

  • bulk_group (Optional[Any]) – Optional grouping key/list for averaging bulk replicates.

  • max_single_cells (int) – Maximum number of reference cells retained for model fitting.

  • top_marker_num (int) – Number of top marker genes used in Bulk2Single preparation.

  • ratio_num (int) – Ratio controlling generated cell numbers per cell type.

  • gpu (Union[int,str]) – Compute device specification (CUDA index, 'mps', or CPU fallback).

Returns:

Initializes bulk-trajectory blending workflow.

Return type:

None

Examples

>>> bulktb = ov.bulk2single.BulkTrajBlend(bulk_seq=bulk, single_seq=adata, celltype_key="celltype")
__init__(bulk_seq, single_seq, celltype_key, bulk_group=None, max_single_cells=5000, top_marker_num=500, ratio_num=1, gpu=0)[source]#

Initialize BulkTrajBlend for trajectory inference and cell blending.

Parameters:

  • bulk_seq (pd.DataFrame) – Bulk RNA-seq matrix with genes as rows and samples as columns.

  • single_seq (anndata.AnnData) – Single-cell reference AnnData used for cell-state prior information.

  • celltype_key (str) – Column in single_seq.obs that stores cell-type annotation.

  • bulk_group (Optional[Any]) – Optional grouping used to aggregate bulk replicates.

  • max_single_cells (int) – Maximum number of single cells used in internal Bulk2Single model.

  • top_marker_num (int) – Number of marker genes per cell type used during preparation.

  • ratio_num (int) – Cell-number ratio used when converting fractions into target counts.

  • gpu (Union[int,str]) – Compute device selector for VAE/GNN workflows.

Returns:

None

Methods

__init__(bulk_seq, single_seq, celltype_key)

Initialize BulkTrajBlend for trajectory inference and cell blending.

bulk_preprocess_lazy()

Preprocess bulk RNA-seq data for trajectory analysis.

gnn_configure([use_rep, neighbor_rep, gpu, ...])

Configure Graph Neural Network for trajectory and transition state analysis.

gnn_generate()

Generate overlapping cell communities representing transition states.

gnn_load(gnn_load_dir[, thresh])

Load a pre-trained GNN model for trajectory analysis.

gnn_train([thresh, gnn_save_dir, gnn_save_name])

Train the GNN model for trajectory and transition state inference.

interpolation(celltype[, adata])

Interpolate trajectory communities back to original data space.

single_preprocess_lazy([target_sum])

Preprocess single-cell reference data for trajectory analysis.

vae_configure([cell_target_num])

Configure the VAE model for bulk-to-single-cell generation.

vae_generate([highly_variable_genes, ...])

Generate trajectory-aware single-cell data with quality filtering.

vae_load(vae_load_dir[, hidden_size])

Load a pre-trained VAE model for trajectory analysis.

vae_train([vae_save_dir, vae_save_name, ...])

Train the VAE model for trajectory-aware single-cell generation.
Contents