molecule

Molecule Objects 

class Molecule()

Molecule class to represent chemical structures using SMILES or InChI. this will include methods for different property calculations and structure comparisons using usearch molecules.

__init__ 

def __init__(name, smiles, fingerprint_dim=2048, radius=2)

Arguments:

  • name:
  • smiles:
  • fingerprint_dim:
  • radius:

def search(library, n=10, metric="tanimoto", using="ecfp4")

Search for similar molecules in a given library using a specified fingerprinting method.

Arguments:

  • library: The dataset to search within.
  • n: Number of similar molecules to return.
  • metric: Similarity metric to use (default is “tanimoto”).
  • using: Fingerprint type to use (default is “ecfp4”).

Returns:

A list of similar molecules from the library.

utils

generate_molecule_dataset 

def generate_molecule_dataset(smiles_files,
                              library_dir,
                              shapes=[shape_maccs, shape_ecfp4, shape_fcfp4],
                              extractor=None,
                              processes=10)

generates a usearch-molecules dataset from a list of molecules and associated fingerprints

along with indicies

Arguments:

  • molecules: list of molecule objects
  • path: where to write the dataset

tanimoto 

def tanimoto(a, b)

computes the tanimoto distance between two boolean numpy arrays

Arguments:

  • a: fingerprint a
  • b: fingerprint b

Returns:

tanimito similarity