deepmedchem 0.2.0b7

Exact RDKit property filters and experimental predicted-property (CP16) acquisition, through one Selection document.

selection = (
    Selection.from_database("enamine-real-v5a")
    .reference("query", smiles="CCOc1ccc(C(=O)N2CCN(C)CC2)cc1")
    .maximize_similarity("rdkit.ecfp4_tanimoto", reference="query")
    .require_preset("lipinski-ro5/v1")
    .where("rdkit.mol_wt", lte=450, units="Da")
    .acquire_predicted_property("openadmet-herg-pchembl", direction="minimize", keep_fraction=0.25)
    .include("properties", "objective_components")
    .limit(100)
)
result = client.selections.create(selection)

• Selection.acquire_predicted_property(...) with typed per-hit (predicted_value, applicable) and response-level (endpoint_id, model_version, direction, units, qualification, before/after counts) acquisition models
• exact assembled-product RDKit values on hits; the server recalculates and enforces every threshold literally
• the same selection contract covers RDKit-filtered random sampling
• compact CLI result tables, database sizes, and vendor price availability in dmc databases
• README, documented examples and the Agent Skill reference updated

CP16 values are predicted, experimental-acquisition-only ranking signals. They may reduce a similarity shortlist; they are not assay results and never establish that an ADMET threshold is met.

Available on Enamine REAL v5a and Freedom Space 5; read the authenticated catalog for each database's properties, presets and endpoints.