UBC Math Colloquium: Josh Scurll

Clustering appears in many guises, playing important roles in diverse areas of cell biology. One such guise is the spatial clustering of proteins on the membrane of a cell. The ability of cell membrane proteins to cluster in response to stimuli is important to the normal function of many cells, but spontaneous, uncontrolled clustering can lead to cancer. Biologists are therefore keen to analyse protein clustering to better understand how cells function and gain insight into related diseases. This quest is assisted by super-resolution microscopy techniques that enable single molecules to be imaged down to nanoscale precision. In this talk, I will outline StormGraph, a graph-based clustering algorithm that I have developed for the analysis of protein clustering in super-resolution microscopy data. Using simulated data, I have found StormGraph to recover ground-truth clusters more accurately than current leading algorithms, and I have demonstrated its use on super-resolution microscopy data from normal and cancerous B-cells, our antibody-producing immune cells.

I will also provide a brief overview of how I intend to use clustering in multi-dimensional proteomic space to potentially improve personalized cancer therapies in the future. Tumours are heterogeneous populations of cells, and the activity of various signalling proteins can differ between cells within the same tumour. This intratumour heterogeneity is a key driver of resistance to cancer therapies, and should therefore be considered if trying to develop effective personalized therapies. I am working to develop suitable experiments and computational analysis to analyse this heterogeneity in B-cell tumours.