SuperPred

SuperPred predicts drug classes and molecular targets by comparing structural fingerprints of query compounds to a reference database to infer ATC classification, targets, and affected pathways.

Key Features:

• Structural Fingerprinting: Converts query molecules into structural fingerprints for comparative analysis.
• Drug Database Comparison: Compares fingerprints against a database of approximately 6,300 drugs to assess structural similarity.
• Database Enrichment: Includes over 7,300 links to molecular targets derived through text mining and manual curation.
• Similarity Metric: Uses the Tanimoto coefficient to quantify structural similarity with a validated threshold above 0.85 for ATC prediction.
• ATC Classification Performance: A Tanimoto threshold >0.85 yields 81% accuracy for WHO Anatomical Therapeutic Chemical (ATC) classification.
• Pathway Annotation: Provides information on affected pathways associated with matched drugs.
• Medical Indication and Lead Identification: Infers likely medical indication areas and potential leads for known targets based on structural similarity.

Scientific Applications:

• Drug classification: Assigns ATC classes to novel compounds based on structural similarity to annotated drugs.
• Target prediction: Suggests molecular targets for compounds via database links obtained by text mining and manual curation.
• Therapeutic hypothesis generation: Proposes potential therapeutic applications and indication areas for novel compounds.
• Lead discovery and repurposing: Identifies new leads for known targets and supports repurposing hypotheses through similarity to existing drugs.

Methodology:

Molecules are converted into structural fingerprints and compared to a database of approximately 6,300 drugs using the Tanimoto coefficient; associations to molecular targets were obtained via text mining and manual curation and pathway information is provided, with a Tanimoto threshold above 0.85 validated to predict WHO ATC classification with 81% accuracy.