AnnData Object Creation

pdb.create_adata

Builds three aligned AnnData objects representing bipartite (Gene-Compound) and unipartite (Gene-Gene, Compound-Compound) networks from pandas DataFrames. All networks are represented as binary adjacency/incidence matrices.

Parameter	Type	Default	Description
gene_compound_df	pd.DataFrame		DataFrame with columns for gene and compound IDs (bipartite relations).
gene_relation_df	pd.DataFrame		DataFrame defining gene-gene relations.
compound_relation_df	pd.DataFrame		DataFrame defining compound-compound relations.
gene_col	str	"symbol"	Column name for gene IDs in gene_compound_df (used for adata_gc obs index).
compound_col	str	"compound"	Column name for compound IDs in gene_compound_df (used for adata_gc var index).
gg_cols	Tuple[str,str]	("symbol1","symbol2")	Column names for the two gene ID columns in gene_relation_df.
cc_cols	Tuple[str,str]	("compound1","compound2")	Column names for the two compound ID columns in compound_relation_df.

Returns

Tuple[ad.AnnData,ad.AnnData,ad.AnnData]: A tuple containing the three aligned AnnData objects:

1. adata_gc: genes×compounds (bipartite incidence matrix).

2. adata_gg: genes×genes (gene-gene adjacency matrix).

3. adata_cc: compounds×compounds (compound-compound adjacency matrix).

Note: The gene index is aligned across adata_gc (obs) and adata_gg (obs/var). The compound index is aligned across adata_gc (var) and adata_cc (obs/var).

Usage

# Assuming pre-processed DataFrames: gc_df, gg_df, cc_df
adata_gc, adata_gg, adata_cc = pdb.create_adata(
    gene_compound_df=gc_df,
    gene_relation_df=gg_df,
    compound_relation_df=cc_df,
    gene_col="symbol",  # Use the mapped column
    compound_col="chebi_id",
)