RDKit

​

Overview

• Chemical structure storage and retrieval
• Substructure and similarity searching
• Chemical structure manipulation
• Molecular descriptor calculation
• Chemical reaction handling

• mol - Represents a molecule (constracted from SMILES notation)
• qmol - Represents a molecule containing query information (constructed from SMARTS notation)
• sfp - Represents a sparse vector fingerprint
• bfp - Represents a bit vectorfingerprint

​

Basic Operations

1. Creating Molecules

-- Create a molecule from SMILES notation
SELECT mol_from_smiles('CCO') AS ethanol;

-- Create a molecule from SMARTS pattern
SELECT qmol_from_smarts('[OH]') AS hydroxyl;

2. Molecular Properties

-- Calculate molecular weight
SELECT mol_amw(mol_from_smiles('CCO')) AS molecular_weight;

-- Count atoms
SELECT mol_numatoms(mol_from_smiles('CCO')) AS atom_count;

3. Substructure Searching

-- Check if a molecule contains a substructure
SELECT mol_from_smiles('CCO') @> qmol_from_smarts('[OH]') AS has_hydroxyl;

​

Similarity Searching

-- Calculate Tanimoto similarity between molecules
SELECT tanimoto_sml(
    morganbv_fp(mol_from_smiles('CCO')),
    morganbv_fp(mol_from_smiles('CCN'))
) AS similarity;

​

Use Cases

• Drug discovery and development
• Chemical database management
• Structure-activity relationship analysis
• Chemical similarity searching
• Reaction prediction and analysis

​

Performance Optimization

1. Create indexes on chemical structure columns:

CREATE INDEX idx_molecule_substructure ON your_table USING gist(molecule);

2. Use appropriate fingerprint types for your specific use case:

• Morgan fingerprints for general similarity searching
• MACCS keys for substructure screening
• Topological fingerprints for specific pattern matching

​

Additional Resources

• RDKit Documentation
• RDKit PostgreSQL Cartridge
• Chemical Development Kit