bio-reaction-enumeration

name: bio-reaction-enumeration description: Enumerates chemical libraries through reaction SMARTS transformations using RDKit. Generates virtual compound libraries from building blocks using defined chemical reactions with product validation. Use when creating combinatorial libraries or enumerating products from synthetic routes. tool_type: python primary_tool: RDKit measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes. allowed-tools: - read_file - run_shell_command

Reaction Enumeration

Generate virtual compound libraries using reaction SMARTS.

Reaction SMARTS Basics

from rdkit import Chem
from rdkit.Chem import AllChem

# Define reaction (reactants >> products with atom mapping)
# Amide coupling: carboxylic acid + amine -> amide
amide_rxn = AllChem.ReactionFromSmarts(
    '[C:1](=[O:2])O.[N:3]>>[C:1](=[O:2])[N:3]'
)

# Validate reaction definition
n_errors = amide_rxn.Validate()
if n_errors[0] == 0:
    print('Reaction is valid')

# Run reaction
acid = Chem.MolFromSmiles('CC(=O)O')
amine = Chem.MolFromSmiles('CCN')

products = amide_rxn.RunReactants((acid, amine))
# products is a tuple of tuples: ((product1,), (product2,), ...)
for prod_set in products:
    for prod in prod_set:
        Chem.SanitizeMol(prod)
        print(Chem.MolToSmiles(prod))

Common Reaction SMARTS

REACTIONS = {
    'amide_coupling': '[C:1](=[O:2])O.[N:3]>>[C:1](=[O:2])[N:3]',
    'reductive_amination': '[C:1]=O.[N:2]>>[C:1][N:2]',
    'suzuki': '[c:1][Br].[c:2][B](O)O>>[c:1][c:2]',
    'buchwald': '[c:1][Br].[N:2]>>[c:1][N:2]',
    'ester_formation': '[C:1](=[O:2])O.[O:3]>>[C:1](=[O:2])[O:3]',
    'michael_addition': '[C:1]=[C:2]C(=O).[C:3]>>[C:1][C:2]([C:3])C(=O)',
}

Combinatorial Library Enumeration

from rdkit import Chem
from rdkit.Chem import AllChem
from itertools import product

def enumerate_library(rxn_smarts, reactant_lists, deduplicate=True):
    '''
    Enumerate products from combinatorial reaction.

    Args:
        rxn_smarts: Reaction SMARTS string
        reactant_lists: List of lists of SMILES for each reactant position
        deduplicate: Remove duplicate products

    Returns:
        List of unique product SMILES
    '''
    rxn = AllChem.ReactionFromSmarts(rxn_smarts)

    # Validate reaction
    if rxn.Validate()[0] != 0:
        raise ValueError('Invalid reaction SMARTS')

    products = []
    seen = set()

    # Generate all combinations
    for reactants in product(*reactant_lists):
        mols = [Chem.MolFromSmiles(s) for s in reactants]
        if None in mols:
            continue

        try:
            prods = rxn.RunReactants(tuple(mols))
            for prod_set in prods:
                for prod in prod_set:
                    try:
                        Chem.SanitizeMol(prod)
                        smiles = Chem.MolToSmiles(prod)

                        if deduplicate:
                            if smiles not in seen:
                                seen.add(smiles)
                                products.append(smiles)
                        else:
                            products.append(smiles)
                    except Exception:
                        continue  # Skip invalid products
        except Exception:
            continue

    return products

# Example: Amide library
acids = ['CC(=O)O', 'c1ccccc1C(=O)O', 'OC(=O)CC(=O)O']
amines = ['CCN', 'c1ccc(N)cc1', 'NCCN']

products = enumerate_library(
    '[C:1](=[O:2])O.[N:3]>>[C:1](=[O:2])[N:3]',
    [acids, amines]
)
print(f'Generated {len(products)} unique products')

Multi-Step Synthesis

def multi_step_enumeration(building_blocks, reaction_sequence):
    '''
    Enumerate products from multi-step synthesis.

    Args:
        building_blocks: Dict of {step: [smiles_list]}
        reaction_sequence: List of reaction SMARTS
    '''
    current = building_blocks[0]

    for step, rxn_smarts in enumerate(reaction_sequence):
        next_bbs = building_blocks.get(step + 1, [])
        if not next_bbs:
            break

        current = enumerate_library(rxn_smarts, [current, next_bbs])
        print(f'Step {step + 1}: {len(current)} intermediates')

    return current

Product Validation

from rdkit import Chem
from rdkit.Chem import AllChem, Descriptors

def validate_products(smiles_list, mw_max=500, remove_reactive=True):
    '''
    Validate and filter enumerated products.
    '''
    valid = []

    reactive_smarts = [
        '[N+]([O-])=O',  # Nitro
        '[Cl,Br,I]',     # Halogens (optional)
        'C#N',           # Nitrile
    ]
    reactive_patterns = [Chem.MolFromSmarts(s) for s in reactive_smarts]

    for smiles in smiles_list:
        mol = Chem.MolFromSmiles(smiles)
        if mol is None:
            continue

        # Check MW
        if Descriptors.MolWt(mol) > mw_max:
            continue

        # Check reactive groups
        if remove_reactive:
            has_reactive = any(mol.HasSubstructMatch(p) for p in reactive_patterns)
            if has_reactive:
                continue

        # Check valence
        try:
            Chem.SanitizeMol(mol)
        except Exception:
            continue

        valid.append(smiles)

    return valid

Reaction Templates

def apply_template(core_smiles, r_groups, attachment_smarts='[*:1]'):
    '''
    Apply R-group decoration to a core scaffold.

    Args:
        core_smiles: Core with attachment point (e.g., '*c1ccccc1')
        r_groups: List of R-group SMILES
        attachment_smarts: SMARTS for attachment point
    '''
    products = []

    for rg in r_groups:
        # Simple string replacement for single attachment
        product_smiles = core_smiles.replace('*', rg, 1)
        mol = Chem.MolFromSmiles(product_smiles)
        if mol:
            try:
                Chem.SanitizeMol(mol)
                products.append(Chem.MolToSmiles(mol))
            except Exception:
                continue

    return products

# Example: Decorate benzene core
core = '*c1ccccc1'
r_groups = ['C', 'CC', 'C(=O)O', 'O']
decorated = apply_template(core, r_groups)

Related Skills

• molecular-io - Save enumerated libraries
• molecular-descriptors - Filter by properties
• admet-prediction - Screen for drug-likeness