liveBook · Manning

Molecular fragment

As explained in Build AI Drug Discovery Pipelines

A molecular fragment is a small chemical structure that can bind a biological target and serve as a starting point or building block for a larger compound.

A molecular fragment is a small chemical structure used as a scaffold, core, or R-group in drug design. Fragments can build compound libraries, probe biological binding sites, or serve as starting points for lead compounds. section 7.2.2 section 7.3

Common computational and medicinal-chemistry uses include:

  • Construct combinatorial libraries by systematically combining predefined scaffolds and R-groups; varying R-groups around a fixed scaffold is scaffold decoration. section 7.2.2
  • Generate alternative chemical series through scaffold hopping, which replaces the scaffold while preserving suitable R-groups and their required orientation. section 7.2.2
  • Support fragment-based drug discovery by screening fragments for target binding, then growing, linking, or merging hits into stronger lead compounds. section 7.3 section 7.3.1
  • Curate molecular datasets by counting disconnected components and excluding molecules with multiple fragments when salts or metals distort measured properties. section 4.1.2

Fragment library and lead generation

A fragment library commonly follows the Rule of 3: molecular weight below 300 Da, no more than three hydrogen-bond donors, no more than three hydrogen-bond acceptors, and CLogP no greater than 3. After screening, a single weakly binding fragment may be extended with R-groups; two non-competitive fragments occupying different sub-pockets may instead be connected with a linker. section 7.3.1

Listing 4.2 Filter training molecules to a single fragment
from rdkit import Chem
from rdkit.Chem import PandasTools
def num_fragments(mol):
    if mol is None:
        return None
    return len(Chem.GetMolFrags(mol))
PandasTools.AddMoleculeColumnToFrame(train_df, 'Drug', 'Molecule')
train_df['fragment_count'] = train_df.Molecule.apply(num_fragments)
train_df = train_df[train_df.fragment_count == 1]    #A

The RDKit listing counts disconnected molecular components and retains only single-fragment molecules, illustrating a practical dataset-curation workflow. section 4.1.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains definitions, Rule-of-3 criteria, library workflows, and RDKit filtering, while deferring detailed binding and optimization methods.
Examples
Uses RDKit component filtering, scaffold decoration, scaffold hopping, Rule-of-3 libraries, and fragment growing, linking, and merging.
What it assumes
Assumes basic molecules and drug-design vocabulary, while defining scaffolds, R-groups, binding sub-pockets, and fragment criteria.
What you can do after
Apply it via RDKit to filter multi-component molecules and design fragment libraries or lead-generation strategies.
How it connects
Connects molecular fragments with scaffolds, R-groups, combinatorial libraries, scaffold hopping, pharmacophore modeling, and fragment-based drug discovery.

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.