Standardization

ChemAudit's standardization pipeline cleans and normalizes chemical structures using a workflow compatible with ChEMBL's curation process. This ensures consistent structure representation across your datasets.

Pipeline Steps

The standardization pipeline consists of four steps, three of which always run:

Step	Description	Always Runs
Checker	Detect structural issues before standardization	Yes
Standardizer	Fix common issues (nitro groups, metals, sulphoxides)	Yes
Get Parent	Extract parent molecule, remove salts and solvents	Yes
Tautomer	Canonicalize tautomers	No (opt-in)

Checker

Identifies potential problems in the input structure:

• Invalid valences
• Unusual bond types
• Suspicious stereo markings
• Radical electrons

Issues are reported but don't stop standardization.

Standardizer

Applies ChEMBL-compatible fixes:

• Normalizes nitro groups
• Standardizes aromatic systems
• Fixes metal disconnections
• Normalizes sulfoxides and related groups
• Fixes rare bond types

Get Parent

Removes salts, solvents, and counterions:

• Identifies and removes common salts (NaCl, HCl, etc.)
• Removes solvents (water, DMSO, etc.)
• Extracts the largest fragment as the parent molecule
• Tracks what was removed

Tautomer Canonicalization (Optional)

Converts tautomers to canonical form:

• Uses ChEMBL's tautomer rules
• Ensures consistent representation
• Warning: May lose E/Z double bond stereochemistry

Stereo Loss

Tautomer canonicalization can change double bond stereochemistry. Only enable if tautomer consistency is more important than preserving E/Z stereo.

How to Standardize

Web Interface

1. Enter your molecule on the Single Validation page
2. Navigate to the Standardization tab
3. View before/after comparison showing:
   • Original and standardized SMILES
   • Steps applied with changes
   • Removed fragments (salts/solvents)
   • Stereo comparison (if changes occurred)
   • Structure comparison (atom count, formula, mass)

API

curl -X POST http://localhost:8001/api/v1/standardize \
  -H "Content-Type: application/json" \
  -d '{
    "molecule": "CC(=O)Oc1ccccc1C(=O)[O-].[Na+]",
    "options": {
      "include_tautomer": false,
      "preserve_stereo": true
    }
  }'

Response:

{
  "result": {
    "original_smiles": "CC(=O)Oc1ccccc1C(=O)[O-].[Na+]",
    "standardized_smiles": "CC(=O)OC1=CC=CC=C1C(O)=O",
    "success": true,
    "steps_applied": [
      {
        "step_name": "standardizer",
        "applied": true,
        "description": "Applied standardization rules",
        "changes": []
      },
      {
        "step_name": "get_parent",
        "applied": true,
        "description": "Extracted parent molecule",
        "changes": ["Removed fragment: [Na+]"]
      }
    ],
    "excluded_fragments": ["[Na+]"],
    "structure_comparison": {
      "original_atom_count": 14,
      "standardized_atom_count": 13,
      "original_formula": "C9H7NaO4",
      "standardized_formula": "C9H8O4",
      "mass_change_percent": -10.87,
      "diff_summary": "Salt removed"
    }
  }
}

Standardization Options

include_tautomer

Type: Boolean Default: false Description: Enable tautomer canonicalization

When enabled, converts tautomers to canonical form using ChEMBL's rules.

Stereo Warning

May lose E/Z double bond stereochemistry. Only enable if tautomer consistency is critical.

preserve_stereo

Type: Boolean Default: true Description: Attempt to preserve stereochemistry during standardization

When enabled, the standardizer tries to maintain R/S and E/Z stereo specifications. However, some transformations may still affect stereochemistry.

Understanding Results

Steps Applied

Each step reports whether it was applied and what changes occurred:

{
  "step_name": "get_parent",
  "applied": true,
  "description": "Extracted parent molecule",
  "changes": ["Removed fragment: [Na+]", "Removed fragment: Cl"]
}

If a step makes no changes, it still reports applied: true but with an empty changes array.

Checker Issues

Problems detected by the checker are reported separately:

{
  "checker_issues": [
    {
      "issue": "unusual_valence",
      "description": "Atom 5 has unusual valence",
      "severity": "warning"
    }
  ]
}

Excluded Fragments

Lists all salts, solvents, and counterions removed:

{
  "excluded_fragments": ["[Na+]", "Cl", "O"]
}

Common excluded fragments:

• Salts: Na+, K+, Cl-, Br-, I-
• Solvents: H2O, DMSO, MeOH, EtOH
• Counterions: TFA-, acetate, formate

Stereo Comparison

If stereochemistry changes, a detailed comparison is provided:

{
  "stereo_comparison": {
    "original_chiral_centers": 2,
    "standardized_chiral_centers": 2,
    "stereo_preserved": true,
    "changes": []
  }
}

Structure Comparison

Shows structural differences between original and standardized:

{
  "structure_comparison": {
    "original_atom_count": 25,
    "standardized_atom_count": 20,
    "original_formula": "C15H20N2O3.HCl",
    "standardized_formula": "C15H20N2O3",
    "original_mw": 312.78,
    "standardized_mw": 276.33,
    "mass_change_percent": -11.65,
    "is_identical": false,
    "diff_summary": "Salt removed"
  }
}

Standardization Provenance

Enable detailed provenance tracking to see exactly what changed at each pipeline stage. This is useful for auditing, debugging, and understanding how your molecule was transformed.

Enabling Provenance

Add include_provenance=true to the API request:

curl -X POST http://localhost:8001/api/v1/standardize \
  -H "Content-Type: application/json" \
  -d '{
    "molecule": "CC(=O)Oc1ccccc1C(=O)[O-].[Na+]",
    "options": {
      "include_provenance": true
    }
  }'

What Provenance Tracks

Each of the four pipeline stages reports detailed changes:

Change Type	What It Records
Charge changes	Atom index, element, before/after charge, normalization rule, SMARTS pattern
Bond changes	Bond index, atom pair, before/after bond type, rule name
Ring changes	Ring atoms, ring size, before/after aromaticity
Radical changes	Atom index, element, before/after radical electrons
Fragment removal	SMILES, fragment name (from 55-entry dictionary), role (salt/solvent/counterion), molecular weight

Tautomer Provenance

When tautomer canonicalization is enabled, provenance includes:

• Canonical tautomer SMILES
• Number of tautomers enumerated
• Modified atoms and bonds
• Complexity flag (set when >100 tautomers were enumerated)
• Whether stereochemistry was stripped

DVAL Cross-References

Provenance records link back to deep validation findings, such as:

• "DVAL-01: 2 undefined stereocenters detected"
• "DVAL-03: 15 tautomers enumerated"

UI: Provenance Timeline

In the web interface, provenance appears as a vertical timeline below the standardization results. Stages with changes auto-expand to show detailed change cards. Stages with no changes remain collapsed.

Use Cases

Database Curation

Standardize structures before adding to databases:

# Batch standardize an entire library
curl -X POST http://localhost:8001/api/v1/batch/upload \
  -F "file=@library.sdf" \
  -F "include_standardization=true"

Structure Deduplication

Standardization helps identify duplicates:

1. Standardize all structures
2. Compare canonical SMILES
3. Group by InChIKey for exact matches
4. Group by Murcko scaffold for structural similarity

ChEMBL Compatibility

If you're comparing against ChEMBL data, standardize using the same pipeline:

curl -X POST http://localhost:8001/api/v1/standardize \
  -d '{"molecule": "...", "options": {"include_tautomer": true}}'

ChEMBL Uses Tautomers

ChEMBL enables tautomer canonicalization by default. Enable it if matching against ChEMBL structures.

Best Practices

1. Always review excluded fragments: Ensure you're not losing important counterions
2. Preserve original structures: Keep both original and standardized for traceability
3. Document options: Record whether tautomer canonicalization was used
4. Check stereo changes: Review stereo comparisons for critical compounds
5. Validate after standardization: Re-validate standardized structures

Limitations

• Salts: Uncommon salts may not be recognized
• Tautomers: Rare tautomeric forms may not be handled correctly
• Stereochemistry: Complex stereo specifications may not preserve perfectly
• Metal complexes: Metal-organic complexes may be over-simplified

Complex Structures

For complex molecules (natural products, metal complexes, macrocycles), always manually review standardization results.

Next Steps

• Database Integrations - Cross-reference standardized structures
• Batch Processing - Standardize large datasets
• API Reference - Full standardization API documentation