Translating Advances in AI/CS to Genomic Discovery

When: Monday, 20 April, 2026 | 18:40-20:00
Where: STEM LAB 1, Gravias 6, The American College of Greece

Co-organized by:
Leonardos Mageiros, Information Technology, Cybersecurity and Computer Science
Ph.D., Microbiology and Infectious Diseases, Swansea University
Deree – The American College of Greece

About the event

AI in genomics is advancing at an unprecedented pace. Current models are now being utilized to predict variant effects, gene regulation, and gene expression, as well as to generate candidate biological sequences. These sophisticated systems include discriminative models, DNA foundation models trained on massive genome corpora, and generative models designed for sequence engineering.

However, the field has reached a critical juncture where the power of these systems is often overstated. While AI can significantly assist in analyzing genomes and proposing candidate designs, it cannot yet reliably create entirely new genomes with predictable, organism-level behavior. Notable limitations remain in the predictability and “real-world” application of these generated sequences.

Safety, Security, and the Rise of “Bioagents”

Beyond technical limitations, the integration of AI and genomics raises significant safety and security concerns. Genomic AI systems may be vulnerable to poisoning through:

• Corrupted data and misleading annotations.
• Biased benchmarks or contaminated knowledge sources.
• Predictions that appear broadly capable but lead to erroneous biological conclusions.

Furthermore, the rise of “bioagents”—systems that connect literature search, sequence design, experiment planning, and interpretation—increases dual-use concerns. These integrated systems may lower the barriers to both beneficial breakthroughs and potentially harmful applications, necessitating a rigorous discussion on the future of genomic security.

About the lecturer

Dr. Georgakopoulos-Soares is an Assistant Professor in the College of Pharmacy at The University of Texas at Austin. His research integrates bioinformatics and AI-driven computational biology to investigate the evolutionary forces shaping genomes, non-canonical DNA structures, and human health and disease, with a particular emphasis on cancer.

A major focus of the lab is the development non-invasive approaches through liquid biopsies, that the group is leveraging to create new methods for early stage cancer detection and identification of neoantigens to develop new cancer therapies in close collaboration with clinical and industry partners. Additionally, mechanistic studies are underway to gain fundamental insights into cancer biology and shape how we approach cancer risk assessment and treatment selection.

Other current projects include the identification of species-specific DNA sequences across hundreds of thousands of genomes and their functional significance in evolution and species differentiation and the characterization of non-canonical DNA structures across large-scale genomic datasets and their effect on gene regulation and genome stability. The group also creates and maintains a variety of freely accessible software tools and databases that serve as resources for the scientific community to accelerate genomic discovery.