AI|ffinity is a lab-in-the-loop* venture company originating from Masaryk
University in the Czech Republic.
DeepHitExplorer, developed by AI|ffinity, is a practical AI-assisted platform
that accelerates hit discovery in the early stages of drug discovery by
predicting ligand epitopes using AI and diversity selection algorithms.
*A research and development method that closely links data processing using AI and simulations with actual experiments (labs), and repeatedly compares AI predictions with real-world experimental results in a loop, dramatically improving the accuracy and efficiency of both.
Click here to contact us.
deepHitExplorer's drug discovery approach
deepHitExplorer is a computational hit discovery platform that harnesses the power of NMR, AI, and cheminformatics.
Using 'explainable AI' and 'ligand-observed NMR experiments' such as STD NMR, WaterLOGSY, and photo-CIDNP, it predicts and presents the complete ligand epitopes of candidate binding molecules. These ligand-observed NMR experiments do not require isotopically labeled proteins and support practical screening workflows, making them attractive techniques for early-stage drug discovery.
- Experiment-integrated AI drug discovery
- deepHitExplorer uses ligand-observed NMR data and explainable AI to predict
the binding affinity and ligand epitope of candidate compounds. By utilizing
NMR experimental information, prediction accuracy is improved compared
to using only binding/non-binding data.
(It can also be used with only binding/non-binding data.) - Compound discovery and diversity selection
- It features an easy-to-use molecular discovery tool and algorithms for selecting a diverse set of compounds, choosing those that exhibit high predictive binding affinity, high solubility, and chemical diversity.
- Applicable to targeting intrinsically disordered protein (IDP)
- Because it is independent of the target protein structure, it can be applied to targets that lack a fixed structure, including intrinsically disordered protein (IDP).
- Optional structure-based design
- When high-quality protein structures are available, you can improve performance by enabling the structure-based scoring module within deepHitExplorer and seamlessly integrating it with ligand-based design tools.
- Standalone application
- It is provided as a standalone application and can be deployed in a typical workstation environment without any special system requirements.
You can watch a video demonstrating this software's approach to discovering hit compounds on Aiffinity's YouTube channel and official website.
Current Recommended System Requirements
- OS
- Linux/Mac/Windows
- Storage
- 50GB
- Memory
- 4GB
- GPU
- NVIDIA (CUDA required)
No specific CPU is recommended. A multi-core processor (4-8 cores) is preferable for parallel processing, but it can also run on a single-core processor.
Why deepHitExplorer?
deepHitExplorer integrates ligand-observed NMR experiments with AI predictions to enable drug discovery that is independent of the structural information of target proteins.