API References

Import CAME:

import came

Example data

load_example_data()

Load example data, for a quick start with CAME.

Pipeline came.pipeline.*

main_for_aligned(adatas, vars_feat[, ...])	Run the main process of CAME (model training), for integrating 2 datasets of aligned features.
main_for_unaligned(adatas, vars_feat, ...[, ...])	Run the main process of CAME (model training), for integrating 2 datasets of unaligned features.
preprocess_aligned(adatas, key_class[, ...])	Packed function for process adatas with aligned features (i.e., one-to-one correspondence).
preprocess_unaligned(adatas, key_class[, ...])	Packed function for process adatas with un-aligned features.
gather_came_results(dpair, trainer, classes, ...)	Packed function for pipeline as follows:

Preprocessing came.pp.*

align_adata_vars(adata1, adata2[, ...])	Align the vaiables of two sc.AnnData objects
normalize_default(adata[, target_sum, copy, ...])	Normalizing datasets with default settings (total-counts normalization followed by log(x+1) transform).
quick_preprocess(adata[, hvgs, ...])	Quick preprocess of the raw data.
quick_pre_vis(adata[, hvgs, normalize_data, ...])	Go through the default pipeline and have a overall visualization of the data.
group_mean(X, labels[, binary, classes, ...])	compute the group averaged features
group_mean_adata(adata, groupby[, features, ...])	Compute averaged feature-values for each group
wrapper_scale(adata[, zero_center, ...])	Wrapper function for centering and scaling data matrix X in sc.AnnData, extended for within-batch processing.
make_bipartite_adj(df_map[, nodes1, nodes2, ...])	Make a bipartite adjacent (sparse) matrix from a pd.DataFrame with two mapping columns.
take_1v1_matches(df[, cols])	Take the one-to-one matches of the given two columns of a pd.DataFrame
subset_matches(df_match, left, right[, ...])	Take a subset of token matches (e.g., gene homologies)
get_homologies(df_match, vals[, cols, ...])	Get the homologous gene of input ones based on the homology-mappings
take_adata_groups(adata, key, group_names[, ...])	Take given groups from an AnnData object
remove_adata_groups(adata, key, group_names)	Remove given groups from an AnnData object
merge_adata_groups(adata, key, group_lists)	Merge the given groups into one single group which is named as '_'.join(groups[i]) by default.
split_adata(adata, key)	Split an AnnData object into a dict of multiple objects

DataPair and AlignedDataPair

make_features(adatas, vars1, vars2, df_varmap)	Decide and make a pair of aligned feature matrices for CAME input.
aligned_datapair_from_adatas(adatas, vars_feat)	Build AlignedDataPair object from a pair of adatas.
datapair_from_adatas(adatas, vars_feat, ...)	Build DataPair object from a pair of adatas.
AlignedDataPair(features, ov_adjs[, ...])	Paired datasets with the aligned features (e.g.
DataPair(features, ov_adjs, vv_adj[, ...])	Paired datasets with the un-aligned features (e.g., cross-speceis)

Graph Neural Network Model

CGGCNet(g_or_canonical_etypes, in_dim_dict)	Cell-Gene-Gene-Cell graph neural network.
CGCNet(**kwargs)	Cell-Gene-Cell graph neural network (used when features are 1-to-1 aligned)

I/O Functions

load_dpair_and_model(dirname[, ...])	Load the output results of CAME.
load_hidden_states(path)	Load hidden states from .h5 file the data structure should be like [ 'layer0/cell', 'layer0/gene', 'layer1/cell', 'layer1/gene', 'layer2/cell', 'layer2/gene' ]
save_hidden_states(data_list, path)	Save hidden states into .h5 file
save_pickle(obj, fpath)	save the object into a .pickle file
load_pickle(fp)	load the object from a .pickle file

Analysis came.ana.*

weight_linked_vars(X, adj[, names, metric, ...])	Computes the similarity of each linked (homologous) pair of variables.
weight_linked_vars_by_expr(dpair, ...)	Compute the weights between homologies by their average expressions across (cell) groups.
make_abstracted_graph(obs_labels1, ...[, ...])	Compute and make the abstracted graph from expression matrices and the linkage weights between homologous genes
compare_modules(mod_labels1, mod_labels2, ...)	Compute common and private genes (cross-species) in each gene module.
compute_common_private(genes1, genes2, gmap)	compute common and private genes based on a given gene mapping

Plotting Functions came.pl.*

plot_stacked_bar(df[, norm, figsize, ...])	helper function for visualizing the group compositions (e.g., in each stage or condition).
heatmap_probas(df_data, lbs[, name_label, ...])	Heatmap of the prediction probabilities
wrapper_heatmap_scores(df_score, obs[, ...])	sort columns and rows, plot heatmap of cell-type scores
plot_confus_mat(y_true, y_pred[, ...])	function for plotting the contingency matrix by default, the values will be normalized by row (true classes)
plot_contingency_mat(y_true, y_pred[, ...])	function for plotting the contingency matrix
embed_with_values(xy, values[, do_zscore, ...])	Visualize the given x-y coordinates, colored by the given (dict of) values
adata_embed_with_values(adata, values[, ...])	This function can also be used to visualize the average expressions of some cell types on gene embedding, in which case, each observation in adata represents a genes.
umap_with_annotates(adt, text_ids[, color, ...])	Plot the UMAP embeddings and annotate the names of the given points
plot_multipartite_graph(g[, subset_key, ...])	display the abstracted multipartite graph (cellType - geneModules - geneModules - cellTypes)