Input Formats

stangene supports two input formats. In both cases, only feature metadata is extracted — the expression matrix is never loaded into memory.

h5ad (AnnData)

The primary format. stangene reads adata.var and adata.var_names to extract feature metadata.

Recognized adata.var columns:

Column	Maps to
gene_ids	original_feature_id (Ensembl ID)
feature_types	original_feature_type (e.g., Gene Expression, Antibody Capture)

When writing results, harmonization columns are added to adata.var in a new *_harmonized.h5ad file. The original var_names are never overwritten.

TSV / CSV

stangene auto-detects common column names:

Detected column name	Maps to
gene, gene_name, feature_name, gene_symbol, symbol	original_feature_name
gene_id, gene_ids, ensembl_id, ensembl_gene_id, feature_id	original_feature_id
feature_types, feature_type	original_feature_type

If your columns have different names, pass an explicit column_map:

ft = stangene.load_features(
    "features.tsv",
    species="human",
    column_map={
        "my_gene_col": "original_feature_name",
        "my_id_col": "original_feature_id",
    },
)

File extension determines the delimiter:

• .tsv, .txt → tab-separated

• .csv → comma-separated