API#

Import HoloNet as

import HoloNet as hn

Preprocessing: `pp`#

Import spatial transcriptomic data into AnnData. Extract and filter the ligand-receptor pair dataframe

`load_brca_visium_10x`()	Load the example dataset
`load_lr_df`([human_or_mouse])	Load the provided dataframe with the information on ligands and receptors.
`get_expressed_lr_df`(interaction_db, ...[, ...])	Filter the dataframe with pairwise ligand and receptor gene, requiring ligand and receptor genes to be expressed in a certain percentage of cells (or spots).

Tools: `tl`#

Some tool functions.

Constructing multi-view CE network#

`elements_expr_df_calculate`(...)	Calculate the expression of elements, including ligand, receptor, co_and_receptor, AG and AN.
`compute_ce_tensor`(used_interaction_db, ...)	Calculate CE matrix for measuring the strength of communication between any pairs of cells, according to the edge weighting function.
`filter_ce_tensor`(ce_tensor, adata, lr_df, ...)	Filter the edge in calculated CE tensor, removing the edges with low specificities.

Computing centralities#

`compute_ce_network_eigenvector_centrality`(...)	Calculate the eigenvector centrality of each cell in the CE network.
`compute_ce_network_degree_centrality`(ce_tensor)	Calculate the degree centrality of each cell in the CE network.

Clustering lr pairs#

cluster_lr_based_on_ce(ce_tensor, adata, ...)

Cluster the LR pairs using the CE network.

Selecting parameters#

default_w_visium(adata[, min_cell_distance, ...])

Calculate a recommended value for the distance parameter in the edge weighting function.

Predicting: `pr`#

Preparing the inputs of the graph neural network#

`get_continuous_cell_type_tensor`(adata[, ...])	Get continuous cell-type information matrix, used as the feature matrix of GNN.
`get_one_hot_cell_type_tensor`(adata[, ...])	Get categorical cell-type labels and one-hot encoded into a matrix, used as the feature matrix of GNN.
`get_gene_expr`(adata[, lr_df, complex_db, ...])	Filter out the genes with too low expression levels or too low dispersions.
`get_one_case_expr`(target_all_expr, ...)	Get a cell_num*1 tensor representing the scaled expression profile of one gene, using as the target of GNN.
`adj_normalize`(adj, cell_type_tensor[, ...])	Process the relationship among cells (such as multi-view CE network or spatial proximity matrix) to form the normalized adjancency matrix as the input of GNN.

Training the graph neural network#

`mgc_repeat_training`(X, adj, target[, ...])	Using cell-type tensor and normalized adjancency matrix as the inputs, repeated training GNN to generate the target gene expression.
`get_mgc_result`(trained_MGC_model_list, X, adj)	Run the trained MGC model and get the generated expression profile of the target gene.
`mgc_training_for_multiple_targets`(X, adj, ...)	Using cell-type tensor and normalized adjancency matrix as the inputs, repeated training GNN to generate the expression of multiple target genes.
`get_mgc_result_for_multiple_targets`(...[, ...])	Run the trained MGC model and get the generated expression profile of target genes.

Model saving and loading#

`save_model_list`(trained_MGC_model_list, ...)	Save the trained model list in model_save_folder/project_name/target_gene_name_list
`load_model_list`(X, adj, project_name[, ...])	Load trained model list in model_save_folder/project_name/

Plotting: `pl`#

Base plotting methods#

`feature_plot`(feature_mat, adata, ...[, ...])	Plot one feature of each cell at the spatial position.
`cell_type_level_network`(sr_network, ...[, ...])	Plot the relationship network (CEs or FCEs) between cell-types.

Plots for spatial datasets#

plot_cell_type_proportion(adata, plot_cell_type)

Plot the proportion of one cell-type in each spot.

Plotting CEs#

`ce_hotspot_plot`(ce_tensor, adata, lr_df, plot_lr)	Plot the centrality of each spot in one LR CE network, representing the hotspot of one LR pair.
`ce_cell_type_network_plot`(ce_tensor, ...[, ...])	Plot the cell-type-level CE network of a certain LR pair.
`lr_umap`(lr_df, cell_cci_centrality[, ...])	Display the LR pair in low-dimention space.
`lr_clustering_dendrogram`(model, lr_df[, ...])	Plot dendrogram plot for the hierarchical clustering model.
`lr_cluster_ce_hotspot_plot`(lr_df, ...[, ...])	Plot the general CE hotspot of each ligand-receptor group.

Plotting FCEs by interpreting the graph neural network#

`lr_rank_in_mgc`(trained_MGC_model_list, lr_df)	Analyze the MGC attention value corresponding to each LR pair, and plot the LR pairs serving as the core mediators of FCEs.
`fce_cell_type_network_plot`(...[, ...])	Display the cell-type-level FCE network of a certain LR pair (or all LR pairs) for a certain target gene.
`delta_e_proportion`(trained_MGC_model_list, ...)	Plotting the proportion of delta_e in the sum of delta_e and e_0 in each cell-type.
`plot_mgc_result`(trained_MGC_model_list, ...)	Plot the generated expression profile of the target gene.

Plots for identifying genes dominated by cell–cell communication#

`find_genes_linked_to_ce`(...[, ...])	Plot the correlation of MGC model results with ground truth, and the results of model only using cell-type with ground truth.
`detect_pathway_related_genes`(...[, fname, ...])	Plotting Heatmaps for specific pathway related genes, and which ligand receptors affect these genes, and these FCEs come from which cell types.

Plots for selecting parameters#

select_w(adata, w_best[, size, alpha, cmap, ...])

Plot the covering spatial region of ligands derived from one spot.

API

Contents