UAB SPARC Showcases Breakthrough in AI-Agent Driven Drug Discovery at Global AI Science Conference

October 22, 2025 — Birmingham, AL. The Systems Pharmacology AI Research Center (SPARC) at the University of Alabama at Birmingham, led by Professor Jake Y. Chen, has achieved a milestone recognition at the 1st Open Conference of AI Agents for Science 2025 hosted online today by Stanford University. Out of more than 350 paper submissions, only 48 were accepted after rigorous peer review by 3 AI reviewers and a human expert each, and among them, UAB SPARC’s work was selected as one of just 11 spotlight presentations — the first and only spotlight representing the drug discovery research domain. The work represents a collaborative effort with SPARC’s Associate Director of Digital Chemistry, Dr. Sixue Zhang, and lead author Fuad Al Abir, a PhD Student from the Department of Biomedical Informatics and Data Science (DBIDS).

The featured study, “Multi-target Parallel Drug Discovery with Multi-agent Orchestration”, demonstrates how a coordinated ecosystem of AI agents can autonomously navigate the early stages of drug discovery—from mining literature and identifying therapeutic targets to generating, evaluating, and optimizing candidate molecules. The framework uses a modular, multi-agent design powered by Google’s Gemini 2.5 Pro, Claude-opus 4.1, and leading generative chemistry models, providing a proof-of-concept blueprint for autonomous discovery pipelines. While NVIDIA’s MolMIM was selected for accessibility in this initial demonstration, future SPARC implementations plan to incorporate better tools to expand predictive and synthetic capabilities.

“Our goal is not just to automate drug discovery but to redefine it,” said ****Prof. Jake Chen, Director of SPARC. “AI agents can explore chemical and biological spaces at unprecedented speed, but expert guidance and quality data remain the keys to success.”

AI Agents at the Frontier of Science

This breakthrough aligns with the global movement led by Dr. James Zou and the Agents for Science Initiative at Stanford, which unites AI leaders and distinguished advisors such as Dr. Eric Topol and chief editors from Nature Biotechnology to chart the next era of “AI scientists.” SPARC’s multi-agent framework exemplifies this vision by integrating autonomous reasoning with scientific transparency, making AI a trusted collaborator in biomedical discovery.

Promise, Challenges, and Future Engagement

SPARC’s framework produced promising drug-like molecules for Alzheimer’s disease targets, including SGLT2, HDAC, and DYRK1A, demonstrating effective multi-target exploration and scaffold hopping from the initial hits as seed compounds in the generative process. However, the study also revealed key limitations—particularly the poor performance of predictive models for data-scarce targets such as CGAS, where limited public datasets constrained accuracy. These findings reinforce that AI-driven drug discovery remains data-dependent and tool-sensitive, underscoring the importance of a human-in-the-loop strategy for model validation and decision-making.

An Open Call to Collaborators

SPARC is actively developing open, transparent, and collaborative AI drug discovery pipelines, welcoming engagement from clinical, translational, and basic biomedical researchers working across the pre-clinical and clinical continuum.

Visit SPARC, housed in the Department of Biomedical Informatics and Data Science at UAB School of Medicine, to learn more about ongoing research, upcoming collaborations, and opportunities to participate in the next generation of AI-empowered biomedical discovery.