Research Projects

This project is aimed at enhancing the adaptability and effectiveness of T-cell receptor (TCR) therapies. TCR therapies are a promising approach to targeting diseases, especially in cancer, because they allow us to precisely target antigens on the surface of cells. However, engineering TCRs has major challenges. TCRs are not very stable, poorly expressed, and most importantly, pose a potential risk for an unwanted immune response. In addition, the process of engineering TCRs to target new antigen epitopes is labor-intensive. We plan to address these issues using an interdisciplinary approach that combines artificial intelligence, computational biology, and wet-lab techniques along with immunology insights to build safer, faster, and more universal TCR-based therapies. The main aim is to denovo design molecules that are soluble bispecific molecules (SBMs) and work like TCRs. This we hope will create a pathway towards the design of more processable SBMs, with the potential for easy tailoring to different diseases. This project has four primary objectives. We will start by creating a multi-parameter TCR/TCRmimic database that contains the sequence, structure and function information of TCRs from different databases. This database will provide the foundational knowledge necessary for effective TCR design. Next, we will use generative AI to create new SBM sequences and structures optimized for stability and efficacy. Our third goal is to develop predictive AI models that can assess SBM expressibility and immunogenicity that will facilitate the screening of potential drug candidates early in the design phase. Lastly, we plan to use the SBMs as a “plug-and-play” framework. This will allow us to swap antigen-binding regions within a stable SBM scaffold, simplifying the process of adapting TCRs to target different antigens. We believe that designed SBMs with the above-discussed qualities will help us create molecules that are stable, non-immunogenic, and customizable. To test this hypothesis, we’ll conduct computational and experimental validations on our designs, feeding each result back into our AI models to improve their predictive accuracy. Achieving these objectives could significantly advance TCR technology. A comprehensive TCR database and our predictive algorithms could become valuable resources for researchers, making TCR design more efficient and cost-effective. The “plug-and-play” framework could make TCR design faster and more adaptable, potentially expanding TCR-based therapies to new diseases. This would also help speed up the development of personalized, adaptable therapies, bringing us closer to a future where TCR-based immunotherapies are tailored to individual patients' needs.