molclaw-mol-similarity

Molecule Similarity Calculation

Invocation

Use the exact tool name and parameter schema exposed by the current harness. Tool names in SDK examples below are raw MCP names, not model-facing aliases. Local files must be uploaded with molclaw-file-transfer before a server tool can use them. Repair PDB inputs first when the workflow requires it.

Scene 1: Compute the Tanimoto similarities between a target molecule and a list of candidate molecules using Morgan fingerprints. Need to use the tool calculate_morgan_fingerprint_similarity.

The description of tool calculate_morgan_fingerprint_similarity.

Compute the Tanimoto similarities between a target molecule and a list of candidate molecules using Morgan fingerprints.
Args:
    target_smiles (str): SMILES string of the target molecule
    candidate_smiles_list (List[str]): List of candidate molecule SMILES strings
    radius (int): Required Morgan fingerprint radius (commonly 2)
    nBits (int): Required Morgan fingerprint vector bit count (commonly 2048)
Return:
    status (str): success/error
    msg (str): message
    similarities (List[dict]): List of dict, each containing the keys 'smiles' and 'score'.
        --smiles (str): A SMILES string of candidate_smiles_list
        --score (float): Similarity value between the candidate SMILES and the target SMILES

How to use tool calculate_morgan_fingerprint_similarity :

response = await client.session.call_tool(
    "calculate_morgan_fingerprint_similarity",
    arguments={
        "target_smiles": target_smiles,
        "candidate_smiles_list": candidate_smiles_list,
        "radius": radius,
        "nBits": nBits
    }
)
result = client.parse_result(response)
similarities = result["similarities"]

Scene 2: Compute the count of shared structural fragments between a target molecule and a list of candidate molecules using Morgan fingerprints. Need to use the tool calculate_common_fragments.

The description of tool calculate_common_fragments.