AI-Driven Discovery of a First-in-Class Candidate for Pancreatic Cancer

Objective: To identify a first-in-class small molecule for pancreatic cancer using an AI-driven drug discovery workflow focused on speed and novelty.

AI-Driven Approach: An AI-powered engine integrated deep learning-based virtual screening, physics-based modeling, and multi-parametric optimization to assess over million compounds generated by Generative AI models.

Candidates were ranked by:

• Predicted target engagement
• Physicochemical and ADMET properties
• Synthetic feasibility and drug-likeness

The top 5 hits were recommended for synthesis.

Synthesis and Evaluation:

2 molecules were synthesized and tested in vitro and in vivo.

One compound (Compound A) showed strong efficacy and was advanced as lead.

Key Preclinical Results:

In Vitro: Compound A was potent across 4 pancreatic cancer lines, including KRAS-mutant and drug-resistant types.

Mechanism: Appears to act via a novel, non-canonical pathway.

In Vivo: In two mouse models, Compound A achieved:

70% tumor growth inhibition

Improved survival

Clean safety profile

PK Profile: Supports once-daily dosing with stable systemic exposure

Current Status: Compound A is in IND-enabling studies, including:

GLP toxicology

Formulation optimization

Preliminary scale-up

Conclusion: This case highlights how AI-driven workflows can accelerate discovery of novel therapies, delivering a promising first-in-class candidate for a challenging oncology target with reduced timelines and resource demands.

Disclaimer: Compound identity, molecular target and specific experimental details withheld due to confidentiality.