Molecule Module

The Molecule module provides lightweight data structures for representing small chemical molecules via SMILES or InChI strings. Built on top of RDKit, it automatically computes molecular descriptors, generates fingerprints (ECFP4, FCFP4, MACCS), calculates Tanimoto similarities, and samples 3D conformers.

User-Facing Methods

Molecule(name, smiles, fingerprint_dim=2048, radius=2): Instantiates a small molecule, generating RDKit Mol objects, 2048-bit ECFP4/FCFP4/MACCS fingerprint bitstrings, InChIKey, and standard molecular descriptors.
similarity(other, fingerprint): Compute the Tanimoto similarity score (float between 0 and 1) between two Molecule instances using "ecfp4", "fcfp4", or "maccs".
generate_conformers(n, prune_thres=0.5, optimize_geom=True): Generate 3D spatial conformers using ETKDGv3 and MMFF geometry optimization. Returns a tuple of (self, conformer_ids).
inchikey(): Calculate and return the IUPAC InChIKey for the chemical structure.

Basic Usage

from benchmate.molecule import Molecule

# Create molecule instances with name and SMILES string
mol1 = Molecule(name="benzene", smiles="C1=CC=CC=C1")
mol2 = Molecule(name="toluene", smiles="Cc1ccccc1")

# Access calculated fingerprints and RDKit molecular descriptors
print(mol1.info.ecfp4)      # Bitstring representation of ECFP4
print(mol1.info.properties) # Dictionary of RDKit descriptors (MW, LogP, TPSA, etc.)
print(mol1.info.inchi)      # InChIKey

# Compute Tanimoto similarity between molecules
sim_score = mol1.similarity(mol2, fingerprint="ecfp4")
print(f"ECFP4 Tanimoto Similarity: {sim_score:.4f}")

# Generate 3D conformers
mol1, conformer_ids = mol1.generate_conformers(n=50, prune_thres=0.5, optimize_geom=True)
print(f"Generated {len(conformer_ids)} conformers")