OdoriFy

OdoriFy predicts interactions between human olfactory receptors (ORs) and odorant molecules using deep neural network-based prediction models combined with explainable artificial intelligence (AI) to characterize agonists, nonagonists, and receptor–ligand pairs.

Key Features:

• Odor Finder: Identifies odorant molecules that interact with wildtype or mutant human olfactory receptors (ORs).
• Odorant Predictor: Classifies input chemicals as odorants or nonodorants using deep neural network-based prediction models.
• OR Finder: Identifies responsive ORs to a given query odorant to map potential receptor–ligand interactions.
• Odorant-OR Pair Analysis: Validates interactions between specific odorants and ORs and provides explainable AI-derived rationales at atomic resolution for odorants and amino-acid-level resolution for ORs.

Scientific Applications:

• Mechanistic studies of olfaction: Characterizes molecular underpinnings of smell perception by identifying OR agonists and nonagonists.
• Receptor–ligand mapping: Maps responsive ORs to odorants to explore the functional relevance of ORs in sensory biology.
• Chemical prioritization for validation: Screens and prioritizes candidate odorants and odorant–OR pairs for experimental follow-up.

Methodology:

Deep neural network-based prediction models trained on a manually curated dataset of human ORs with known agonists and nonagonists, coupled with explainable AI that provides atomic-resolution rationales for odorant/nonodorant predictions and amino-acid-level rationales for OR predictions.