AI + Immunology
AI-Guided Classification of Chronically Hepatitis B Virus-infected Patients
Integration of HBV-specific T cell immune profiles with virological and biochemical analysis to stratify patients to guide the clinical management of chronic HBV patients
Currently, clinical classification and management of chronic HBV (CHB) patients relies exclusively on the assessment of virological and biochemical biomarkers despite the importance of HBV-specific T cells in viral control [1]. We are developing a robust and rapid whole blood assay (WBA) [2] that allows the measurement of the functional secretome of HBV-specific T cells in large patient populations. We will deploy AI algorithms to separate CHB patients within similar clinical phases into clusters with distinct HBV-specific immune profiles. This AI-guided classification of patients within similar clinical disease phases, based on their antiviral T cell function, will provide a novel biomarker for interpreting host-viral interactions and can signpost the selection of novel immunotherapies.
[1] Bertoletti, A. & Le Bert, N. Quest for immunological biomarkers in the management of CHB patients. Gut (2023). https://doi.org:10.1136/gutjnl-2023-329437
[2] Tan, A. T. et al. Rapid measurement of SARS-CoV-2 spike T cells in whole blood from vaccinated and naturally infected individuals. J Clin Invest 131 (2021). https://doi.org:10.1172/JCI152379
Leveraging Deep-Learning Generative Models for Virtual Screening of Drug Perturbations to Combat Drug Resistance and Reverse Immune Aging
In the era of abundant single-cell data, the utilization of deep-learning based generative models emerges as a promising avenue. These models offer the potential to serve as in silico representations of cellular systems, analogous to traditional mouse models or in vitro setups. Importantly, they provide a computational platform where perturbations can be introduced to simulate various conditions. This computational framework enables the efficient virtual screening of a vast array of perturbations at minimal cost, serving as an initial step to identify promising targets for further experimental validation. In this context, our focus lies particularly on modelling drug perturbations to elucidate the effects of small molecules across diverse cell types. Such an approach holds significant promise in combating drug resistance in cancer and addressing age-related immune phenotypes. By leveraging deep-learning generative models, we aim to facilitate a paradigm shift towards more efficient drug discovery and therapeutic interventions.