AI-driven Drug Discovery Platform

Inventum.AI's AI-driven Drug Discovery Platform is a generative AI-powered solution designed to accelerate the drug discovery process, reduce costs, and enhance the efficiency of identifying novel drug-like molecules. The platform is built for researchers and organisations in the life sciences sector seeking to streamline the journey from protein target identification to candidate compound generation.

At the core of the platform is the Nature Based Generator (NBG) approach, a structured workflow for designing novel ligands for protein targets. This methodology combines binding site identification, grid construction, scaffold generation, and periphery elaboration to systematically produce and evaluate potential drug candidates.

Generation of New Ligands for a Given Protein

• Identification of ligand-protein binding sites using the proprietary SiteRadar algorithm
• Molecular docking to assess how candidate molecules interact with target proteins
• De novo structure generation for entirely new molecular scaffolds
• Scaffold growing to expand and elaborate on core chemical structures
• Intermolecular interactions analysis to understand binding behaviour at the atomic level

Annotation of Generated Compounds

• Prediction of binding affinity to evaluate how strongly a ligand interacts with its target
• Evaluation of ADMET properties (Absorption, Distribution, Metabolism, Excretion, and Toxicity) to assess drug-likeness and safety profiles

Platform Workflow Stages

1. Site Radar: Prediction of the active site directly from the protein's 3D structure
2. Site Selection: Binding site selection informed by known ligands
3. Site Map: Arrangement of probes across the binding site to map interaction opportunities
4. Scaffold Generation: Generation of a core scaffold structure based on the probe data
5. Scaffold Placement: Positioning of a given 2D scaffold within the binding site
6. Periphery Generation: Generation of substituents around the selected scaffold to produce diverse candidate structures

Key Technical Advantages

• SiteRadar demonstrates higher accuracy in binding site identification compared to established algorithms such as Fpocket and PUResNet
• Accurate classification of ligand-protein binding sites
• High ability to predict the binding mode of candidate ligands
• Strong correlation with experimental ligand-protein affinity data
• Precise ability to distinguish between positions of different atom types within binding sites

Case Studies

• AI-driven discovery of an IRAK4 inhibitor (published December 2024)
• Design and validation of FLT3 inhibitors for Acute Myeloid Leukemia (AML) treatment (published January 2025)

Inventum.AI's platform is accessible for direct trial, enabling research teams to explore its generative AI capabilities hands-on. The combination of proprietary algorithms, a structured multi-stage workflow, and compound annotation tools positions the platform as a comprehensive end-to-end solution for modern computational drug discovery.