liveBook · Manning

Combinatorial library design

As explained in Build AI Drug Discovery Pipelines

Combinatorial library design is the systematic construction of many compounds by combining predefined molecular fragments under specified chemical constraints.

Combinatorial library design constructs many unique compounds by systematically combining predefined molecular fragments. A fragment may be a scaffold, or core, or an R-group attached to that scaffold. Keeping the scaffold fixed while varying R-groups creates analogs for exploring how structural changes affect activity, such as through a QSAR model. section 7.2.2

A practical design workflow can:

  • Extract Murcko scaffolds and R-groups from molecular structures. section 7.2.2
  • Combine fragments through reaction rules while limiting expansion and retaining chemically valid molecules. section 7.2.2
  • Use scaffold decoration for optimization, or scaffold hopping when the original core has toxicity, patentability, or other fundamental limitations. section 7.2.2
Listing 7.5 Combinatorial Library Construction
from rdkit.Chem.Scaffolds import MurckoScaffold
from rdkit.Chem import AllChem
import random
 
cores, frags = [], []
data['mols'] = data['SMILES'].apply(Chem.MolFromSmiles)
for mol in data['mols']:
    if mol is None:
        continue
 
    bms = MurckoScaffold.GetScaffoldForMol(mol)    #A
    core = Chem.ReplaceSidechains(mol, bms)
    rgroups = Chem.ReplaceCore(mol, bms)    #A
    if core is None:
        continue
    cores.append(Chem.MolToSmiles(core))
    if rgroups:
        frags.extend(Chem.MolToSmiles(rgroups).split('.'))
 
reaction_smarts = "[*:1][#0].[#0][*:2]>>[*:1][*:2]"
rxn = AllChem.ReactionFromSmarts(reaction_smarts)
 
new_mols = []
EXPANSION_FACTOR = 5
MAX_STEPS = 20
for core_smi in cores:    #B
    for _ in range(EXPANSION_FACTOR):
        current_smi = core_smi
        steps = 0
        while "[*]" in current_smi and steps < MAX_STEPS:
            steps += 1
            random_index = random.randint(0, len(frags) - 1)
            products = rxn.RunReactants((
                Chem.MolFromSmiles(current_smi),
                Chem.MolFromSmiles(frags[random_index]),
            ))
            if len(products) == 0:    #C
                break    #C
            product_index = random.randint(0, len(products) - 1)
            current_smi = Chem.MolToSmiles(products[product_index][0])
        mol = Chem.MolFromSmiles(current_smi)
        if mol:
            AllChem.Compute2DCoords(mol)
            new_mols.append(mol)

The RDKit listing demonstrates scaffold extraction, R-group collection, reaction-based assembly, expansion controls, and molecule validation. section 7.2.2

Figure 7.8 Scaffold hopping requires identifying a new scaffold that can bear the required R-groups in the appropriate relative positions. Replacement of the benzene ring with a pyridine ring would give a similar structure, but a pyrimidine scaffold would not work since the two nitrogens won’t accommodate the necessary R-group orientation.

The scaffold-hopping figure shows why replacement cores must preserve the required R-group orientation: pyridine can work where pyrimidine cannot. section 7.2.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains fragment roles, scaffold decoration, scaffold hopping, and RDKit construction, but does not detail library-diversity selection.
Examples
Uses RDKit scaffold extraction and reaction assembly, plus a pyridine-versus-pyrimidine scaffold-hopping comparison.
What you can do after
Build it from scratch with Python and RDKit by extracting fragments, assembling candidates, limiting expansion, and sanitizing molecules.
How it connects
Connects molecular scaffolds, R-groups, scaffold decoration, scaffold hopping, QSAR models, retrosynthetic library design, and clustering.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.