plinder.core.scores

The plinder.core.scores subpackage provides a consistent API for querying the various parquet collections in the PLINDER dataset. The preferred parquet reader engine is duckdb, but much of the code previously used pandas and pyarrow directly. The internal query API supports converting the same pyarrow query filters used in pd.read_parquet into raw SQL for duckdb to execute.

query_ligand_similarity	Query the ligand similarity database and return the results.
cross_ligand_similarity	Query the ligand similarity database for a cross similarity between a set of query and target ligands.
query_protein_similarity	Query the protein similarity database for a given search_db and return the results.
cross_protein_similarity
query_clusters	Query the cluster database.
query_links	Query the linked systems dataset
query_index	Query the index database.