---

name: reactions-standardization description: "Use for RDKit reaction SMARTS/RXN workflows, product sanitization, MolStandardize cleanup/normalization/fragment/tautomer handling, R-group decomposition, stereochemistry/CIP practical workflows, and medicinal chemistry transformations. Route base molecule parsing to molecule-io-core and optional MMPA/Fraggle contrib workflows to contrib-utilities." disable-model-invocation: true metadata: disco-role: operating license: BSD 3-Clause

RDKit Reactions and Standardization

Use this sub-skill when a task asks an agent to transform molecules with reaction SMARTS, clean medicinal chemistry structures, choose parent fragments/charge forms, enumerate or canonicalize tautomers, decompose analog series into R-groups, or preserve/debug stereochemistry through these workflows.

Route here

• Build and run reaction SMARTS/RXN workflows with rdkit.Chem.rdChemReactions or AllChem.ReactionFromSmarts.
• Validate reaction definitions, atom mapping, reactant counts, agents, product templates, and product sanitization.
• Use rdkit.Chem.MolStandardize.rdMolStandardize for cleanup, normalization, reionization, uncharging, fragment parents, charge parents, and tautomer enumeration.
• Run rdkit.Chem.rdRGroupDecomposition.RGroupDecompose against labeled or auto-labeled cores and interpret unmatched molecules.
• Preserve, assign, or inspect stereochemistry and CIP labels after transformations with Chem.FindPotentialStereo and rdCIPLabeler.AssignCIPLabels.
• Implement medicinal chemistry transformations such as neutralization, salt stripping, parent selection, scaffold analog decomposition, and small reaction-based substitutions.

Route elsewhere

• Base molecule parsing, suppliers, None checks, and generic sanitization basics: molecule-io-core.
• Descriptors, fingerprints, similarity, and clustering after standardization: descriptors-fingerprints.
• Drawing molecules or reactions and coordinate generation: conformers-drawing.
• Optional contributed MMPA, Fraggle, SA/NP scoring, and other Contrib/ utilities: contrib-utilities.
• RDKit source checkout build/test work for these modules: repo-development.

Start with these references

• references/reactions.md for reaction SMARTS construction, running reactions, product handling, and stereochemistry behavior in reactions.
• references/standardization-rgroups-stereo.md for MolStandardize, R-group decomposition, tautomer, uncharging, fragment-parent, and CIP/stereo recipes.
• references/troubleshooting.md for invalid SMARTS, unsanitized products, unmatched cores, and parameter mistakes.
• scripts/standardize_react_smoke.py for a tiny standalone cleanup plus reaction SMARTS smoke test.

Core workflow

1. Parse molecules in molecule-io-core, then pass checked Mol objects into reaction or standardization code.
2. For reaction SMARTS, build the reaction, inspect template counts, call Validate(), and match the exact reactant tuple arity required by the reaction.
3. Treat RunReactants() output as unsanitized candidate products: copy or select products deliberately, run Chem.SanitizeMol, and report failures with the product index and SMILES when possible.
4. Standardize before comparing analogs or calculating descriptors: choose whether the task needs Cleanup, FragmentParent, ChargeParent, Uncharger, or tautomer canonicalization rather than applying every transform blindly.
5. For R-group decomposition, start with a chemically meaningful core, prefer labeled attachment points when labels matter, and always inspect unmatched indices before trusting the R-group table.
6. For stereochemistry-sensitive workflows, keep isomericSmiles=True, use mapped reaction atoms, assign CIP labels after final sanitization, and document whether a transform preserves, creates, destroys, or inverts a stereocenter.

Minimal examples

from rdkit import Chem
from rdkit.Chem import rdChemReactions

rxn = rdChemReactions.ReactionFromSmarts("[C:1]=[O:2]>>[C:1][O:2]")
products = rxn.RunReactants((Chem.MolFromSmiles("CC=O"),))
product = products[0][0]
Chem.SanitizeMol(product)
smiles = Chem.MolToSmiles(product, isomericSmiles=True)

from rdkit import Chem
from rdkit.Chem.MolStandardize import rdMolStandardize

mol = Chem.MolFromSmiles("CC(=O)[O-].[Na+]")
parent = rdMolStandardize.FragmentParent(mol)
uncharged = rdMolStandardize.Uncharger().uncharge(parent)

Bundled check

Run the bundled helper in an environment where RDKit is importable:

python scripts/standardize_react_smoke.py --smiles "CC(=O)[O-].[Na+]" --reactant "CC=O"

It asserts that cleanup and fragment-parent selection produce valid molecules, builds a tiny reaction SMARTS, sanitizes the first product, and prints canonical SMILES outputs. Use --bad-reaction to confirm invalid reaction SMARTS are reported cleanly.