Flora (Yimeng) Liu, University of Victoria

Co-Supervisors – Farouk S. Nathoo (UVic) and Brad Nelson (UBC, BC Cancer)

Collaborators – Celine Laumont (BC Cancer), Shreena Kalaria (BC Cancer, UVic)

Project Description: The project aims to develop a statistical approach to infer individual heavy and light chain/ α and β chain pairs of B cell and T cell clonal families from spatial transcriptomic data. This will
utilize gene expression data as well as spatial co-expression point patterns. We propose an
innovative method that leverages combinatorial optimization by utilizing an estimated mapping
matrix, derived from the normalized expression matrices for both chain types, and the
distances between spatial point patterns. The best pairs can be obtained through optimizing an
objective function that combines the gene expression mapping matrix and the spatial distance
matrix in a way that is useful for understanding the antigens recognized by B cells and T cells in
ovarian cancer. The statistical significance of the estimated clone pairs will be based on
permutation-based approaches applied to the solution. Simulation studies will be conducted to
investigate the importance of accounting for spatial information by simulating point patterns
with varying distances between heavy and light chains. The project is a collaborative effort
involving the Department of Mathematics and Statistics at the University of Victoria and the
Deeley Research Centre of BC Cancer. This collaborative effort aims to better understand the
hottest immune neighbourhoods in ovarian cancer, which is critical for advancing
immunotherapy.