UCLA

Antibodies are essential to adaptive immunity and therapeutic development. IgG antibodies coordinate immune effector responses by selectively binding to target antigens and interacting with various effector cells via Fcγ receptors. In this study, I explore two computational strategies for modeling antibody responses. First, I extend and employ a mechanistic model to analyze mixed Fc IgG binding measurements. This multivalent binding model efficiently predicts interactions between mixtures of multiple multivalent ligands and multiple cell surface receptors. Applied to experimental data, this model quantitatively matches mixed FcγR binding measurements, refines affinity estimates, and predicts antibody-mediated immune effector cell responses. Notably, it highlights IgG2’s binding capabilities to FcγRI, contrary to previous nonbinding estimations. Second, I adopt a data-driven approach using tensor-based methods to deconvolute systems serology data. Given the complexity of recent biological research characterized by measurements in multiple degrees of variation, I provide an overview of applying tensor methods to high-throughput biological datasets. Applied these principles to HIV- and SARS-CoV-2- infected patients’ serum sample data, tensor methods reveal consistent patterns and outperform traditional methods in data reduction and prediction accuracy, emphasizing their efficacy in identifying immune functional responses and disease status. Overall, this study demonstrates how mechanistic and data-driven approaches can be effectively applied to analyze antibody-mediated immunity, showcasing their distinct roles in computational biology.