AIMEE

AIMEE identifies inhibitors of the 3C-like protease (3CLpro) of SARS-CoV-2 by combining deep learning-derived predictions with experimental validation to prioritize bioactive compounds for antiviral development.

Key Features:

• Integration of AI and Experimental Approaches: AIMEE combines deep learning-based predictions with enzymological experimental validation to iterate between computational selection and biochemical testing.
• High-Throughput Screening Capability: Using a bioactive chemical library, AIMEE conducted high-throughput screening that identified six novel inhibitors against 3CLpro with a hit rate of 29.41%.
• Interpretability and Domain Knowledge Mapping: Deep learning-extracted features are mapped to chemical properties from domain knowledge to support interpretation of model predictions and compound selection.
• Activity-Based Probing: The framework identified a commercially available compound that functions as an activity-based probe for 3CLpro.

Scientific Applications:

• Drug Discovery and Development: Identification of novel SARS-CoV-2 3CLpro inhibitors with low IC50 values (e.g., <3 μM) supports antiviral lead identification and optimization.
• Protein-Ligand Interaction Studies: The framework aids investigation of protein–ligand interactions relevant to 3CLpro inhibition and mechanistic enzymology.

Methodology:

The deep learning model was trained to predict potential 3CLpro inhibitors by analyzing chemical libraries and extracting features that correlate with inhibitory activity against 3CLpro.