This presentation explores cutting-edge artificial intelligence (AI) approaches for the rational design of nucleic acid therapeutics, including linear mRNA, circular RNA (circRNA), and small interfering RNA (siRNA). Key topics include the application of generative AI and large language models for novel drug candidate design and mRNA druggability prediction, aiming to enhance efficiency and significantly reduce the cost of biological experiments.

SyzOnc is an integrated oncology platform and therapeutics company developing next generation treatments for solid tumor patients. Our therapies will provide patients with new approaches and clinicians with new tools to fight devastating solid tumors.

Most people diagnosed with solid tumors (like sarcomas or liver cancer) have few effective treatment options. The standard of care is currently non-curative. These cancers quickly evolve resistance to today’s standard treatments, including chemotherapy and immunotherapy. The result? Too many patients face relapse and limited survival. The reason is simple but overlooked: cancer isn’t just about cancer cells. Tumors are ecosystems made up of cancer cells, immune cells, and a protective physical infrastructure called the extracellular matrix (ECM). These elements work together to help the tumor grow, evade the immune system, and block treatments.

Choosing anti-cancer molecular targets based on oncogenic driver mutations, or a singular function such as “ability to enhance proliferation”, invites tumors to evolve workarounds that make these point solutions fail. Tumors are not static, they evolve, and we need solutions that are powerful enough to eliminate their ability to rebound. Furthermore, immuno-oncology fails for many cancer types because it is ineffective in solid tumor microenvironments where the extracellular matrix is tumor protective. Our approach to solid tumors uses AI/ML tools to target critical features of tumor biology untapped by the traditional pharma pipeline, specifically the 3D matrix-rich tumor ecosystem. Our unique next generation targets and rapid phenotypic drug screening approach will bring clinical success to solid tumor patients.

We developed a platform to eliminate the bench to bedside failures in target discovery that have plagued oncology. It integrates public and proprietary tumor ecosystem data encompassing several key aspects of tumor biology: cancer cells, matrix, and immune cells. This approach will reveal novel “triple threat” targets that control all three areas, simultaneously. Our first small molecules are under development and we look forward to delivering them to solid tumor patients.

The rapid advancement of AI models in recent years has transformed the field of protein design, enabling the creation of de novo proteins with precisely defined structural and functional properties. These innovations are now being applied to overcome limitations of conventional antibodies, such as their large size, limited tissue penetration, and restricted targeting capabilities. Using state-of-the-art generative AI models, ZipBio has developed the COMPACT platform—a rapid, highly efficient tool for generating small, modular, potent, and highly selective de novo proteins for diseases with significant unmet medical need. Leveraging this platform, we have designed small binders capable of engaging multiple targets on a single interface, as well as molecules with conditionally activated binding for precise, more effective and potentially safer interventions, illustrating the platform’s versatility in creating next-generation therapeutics.

This presentation demonstrates how AI-integrated platforms are transforming therapeutic development through three commercial solutions at Bioelectronica:

- Hypercell Platform® combines computer vision with AI-powered single-cell analysis for next-generation antibody discovery through intelligent cell sorting and real-time analytics. Over 90+ discovery campaigns have demonstrated significant reduction in lead identification timelines while improving candidate quality through AI-guided selection.

- AbFinder Software™ leverages machine learning for antibody sequence optimization and developability prediction in drug discovery pipelines. This cloud-based platform enables data-driven therapeutic antibody hit selection, reducing late-stage failures through predictive analytics and automated candidate ranking for pharmaceutical applications.

- SmartCulture X™ applies AI-optimized imaging & automation to microbiome screening for live biotherapeutics development. The platform supports next-generation microbiome-based treatments through intelligent strain isolation and characterization workflows.

These integrated platforms showcase quantifiable improvements in drug discovery efficiency, from months to weeks in lead identification, while enabling previously impossible therapeutic modalities through AI-enhanced detection capabilities.

Our research focuses on two areas for precision gene and cell engineering: new tool development and AI-enabled engineering. First, we are moving beyond the current limitations of CRISPR-Cas9 by focusing on new enzymes for large, no-DNA-damage gene insertions. We have engineered single-strand annealing proteins (SSAPs) for cleavage-free knock-in and are excited to collaboratively apply these tools in ex vivo and animal models. Second, to accelerate the pace of gene and cell engineering, our lab is developing AI agents for biomedical research to automate gene editing, cell engineering, and genomic discovery, from hypothesis generation to data interpretation. This AI-native approach has shown significant potential in areas such as CRISPR-GPT for gene-editing, RNAGenesis for therapeutic RNA design, and in stem cell engineering and functional screening (AutoScreen, Stella), showcased at genomics.stanford.edu. Our work demonstrates the broad utility of new tools, AI models and agents in accelerating genome and cell engineering.