Configuration & Loader

Shared loader for mockdock benchmark config and bioactivity data. Used by both MDOracle and MDEvaluator.

class mockdock.loader.BenchmarkLoader(benchmark_name, scratch_dir=None)

Bases: object

Lightweight loader: reads the benchmark TOML config and bioactivity CSV. Does NOT initialise any docking engine or create run directories.

Parameters:

• benchmark_name (str)

• scratch_dir (Path | None)

classmethod list_benchmarks()

Return type:

list[str]

get_full_data_and_threshold()

Load bioactivity data and compute the 25th-percentile activity threshold.

Lookup order: 1. In-memory cache 2. Package-bundled CSV (mockdock/bioactivity_data/<name>.csv) 3. Scratch cache (~/.mockdock/bioactivity_data/<name>_chembl.csv) 4. Live ChEMBL fetch

Return type:

tuple[DataFrame, float, str]

get_initial_compounds()

Return the model-visible initial compound set (lowest-quartile bioactivity).

Return type:

DataFrame

get_validation_compounds()

Return the validation compound set (above the 25th-percentile threshold).

Return type:

DataFrame