Structural Alerts

Structural alerts identify potentially problematic substructures in molecules. ChemAudit screens against over 1,500 patterns from industry-standard catalogs including PAINS, BRENK, NIH, ZINC, and ChEMBL filters.

Available Alert Catalogs

PAINS (Pan-Assay Interference Compounds)

Patterns	Severity	Description
~480	Warning	Compounds that interfere with multiple assay types

PAINS includes three subcatalogs:

• PAINS_A: Most frequent interference patterns
• PAINS_B: Moderate interference patterns
• PAINS_C: Less frequent but still problematic patterns

FDA-Approved Drugs

87 FDA-approved drugs contain PAINS patterns. Alerts are warnings for investigation, not automatic rejections.

BRENK

Patterns	Severity	Description
~105	Warning	Unwanted chemical moieties

Filters for functional groups and substructures associated with poor pharmacokinetics, toxicity, or reactivity.

NIH

Patterns	Severity	Description
~180	Warning	NIH MLSMR excluded structures

Patterns excluded from the NIH Molecular Libraries Small Molecule Repository screening collection.

ZINC

Patterns	Severity	Description
~95	Warning	ZINC database filters

Filters used by the ZINC database to identify problematic compounds.

ChEMBL Pharma Filters

Patterns	Severity	Description
~700+	Warning	Combined from 7 pharmaceutical industry sources

ChEMBL structural alerts are curated from real-world pharmaceutical screening experience:

Sub-Catalog	Source	Focus
BMS	Bristol-Myers Squibb	Reactive functional groups — electrophiles, Michael acceptors, acyl halides, and other chemically reactive moieties that cause non-specific binding
Dundee	University of Dundee	Promiscuous compound filters — patterns frequently appearing as false positives across diverse target screens
Glaxo	GlaxoSmithKline	Undesirable moieties — toxicophores, metabolically labile groups, and structural features associated with clinical failures
Inpharmatica	Inpharmatica Ltd.	Chemical liabilities — ADMET flags and functional groups with known pharmacokinetic issues
LINT	Lead Identification Noise	Noise-causing patterns in lead identification — substructures that generate artifacts in screening cascades
MLSMR	NIH/MLSMR	Molecular Libraries Screening Center Network — additional exclusion patterns beyond the standard NIH set
SureChEMBL	EMBL-EBI	Patent literature structural alerts — patterns flagged from analysis of chemical patent claims

When ChEMBL alerts are enabled, all 7 sub-catalogs are screened simultaneously. Each matched alert reports which sub-catalog it belongs to.

How to Screen

Web Interface

1. Enter your molecule on the Single Validation page
2. Navigate to the Structural Alerts tab
3. View matched alerts with:
   • Pattern name and description
   • Severity level
   • Matched atoms (highlighted in structure)
   • Catalog source

API - Full Screening

Get detailed information about all matched alerts:

curl -X POST http://localhost:8001/api/v1/alerts \
  -H "Content-Type: application/json" \
  -d '{
    "molecule": "c1ccc2c(c1)nc(n2)Sc3nnnn3C",
    "catalogs": ["PAINS", "BRENK"]
  }'

Response:

{
  "status": "completed",
  "alerts": [
    {
      "pattern_name": "thiazole_amine_A",
      "description": "Potential assay interference",
      "severity": "warning",
      "matched_atoms": [4, 5, 6, 7],
      "catalog_source": "PAINS_A",
      "smarts": "[#7]-c1nc2ccccc2s1"
    }
  ],
  "total_alerts": 1,
  "screened_catalogs": ["PAINS_A", "PAINS_B", "PAINS_C", "BRENK"],
  "has_critical": false,
  "has_warning": true
}

API - Quick Check

Fast yes/no check without detailed results:

curl -X POST http://localhost:8001/api/v1/alerts/quick-check \
  -H "Content-Type: application/json" \
  -d '{
    "molecule": "c1ccc2c(c1)nc(n2)Sc3nnnn3C",
    "catalogs": ["PAINS"]
  }'

Response:

{
  "has_alerts": true,
  "checked_catalogs": ["PAINS"]
}

Performance

Use quick-check for large-scale filtering, then run full screening on flagged molecules for details.

Interpreting Results

Severity Levels

Severity	Meaning	Action
Warning	Potentially problematic	Investigate further, review literature
Critical	Definitively problematic	Avoid or have strong justification

Most structural alerts are classified as warnings because context matters. A PAINS alert in a screening library compound warrants investigation, but the same pattern in a clinical candidate with supporting data may be acceptable.

Matched Atoms

The matched_atoms field indicates which atoms in the molecule match the alert pattern. In the web interface, these atoms are highlighted in the structure viewer.

This helps you:

• Understand which part of the molecule triggered the alert
• Identify structural modifications to avoid the alert
• Assess whether the match is a false positive

False Positives

Structural alert matching is based on SMARTS patterns, which can sometimes match unintended substructures. Always review matched atoms to verify the alert is relevant.

Using Alerts in Your Workflow

Early-Stage Screening

Screen compound libraries before purchasing or synthesizing:

# Batch process with alerts enabled
curl -X POST http://localhost:8001/api/v1/batch/upload \
  -F "file=@library.sdf" \
  -F "include_extended_safety=true" \
  -F "include_chembl_alerts=true"

Filter results to exclude compounds with alerts, or flag them for manual review.

Compound Prioritization

Use alert results to prioritize compounds:

1. No alerts: High priority for screening
2. PAINS only: Medium priority, review assay compatibility
3. Multiple catalogs: Low priority, requires justification

Hit Validation

When validating screening hits, check for alerts:

• Hits with PAINS alerts may be assay artifacts
• Structural modifications that remove alerts often improve hit quality
• Literature search for the alert pattern provides context

Alert Statistics in Batch Results

Batch processing provides alert summary statistics:

{
  "alert_summary": {
    "PAINS": 12,
    "BRENK": 8,
    "NIH": 3,
    "ZINC": 5,
    "ChEMBL": 15
  }
}

Use this to:

• Assess overall library quality
• Identify prevalent problematic patterns
• Guide library cleanup efforts

Catalog Selection

Default Screening

By default, ChemAudit screens against PAINS only. This provides a good balance of speed and coverage.

Extended Safety

Enable extended safety to include NIH and ZINC filters. Use for:

• Compound library curation
• Pre-purchase screening
• Academic screening collections

ChEMBL Alerts

Enable ChEMBL alerts for comprehensive pharma-industry filters. Use for:

• Drug discovery projects
• Lead optimization
• Industrial screening

Performance Impact

Screening all catalogs takes longer. For large batches, start with PAINS, then re-screen flagged compounds with full catalogs.

Next Steps

• Scoring Overview - Comprehensive molecular scoring
• Safety Filters - Safety scoring vs. alerts
• Batch Processing - Screen large libraries
• API Reference - Full alerts API documentation