Modelling cell-cell interactions with random graphs: a journey from biology to math and back.

Link to Join Zoom Meeting

**This talk will be broadcast at 13:30 BT / 14:30 CET / 15:30 EET, May 27th, 2026 on Zoom only. **

Meeting-ID: 667 4776 1513 Passcode: 834059

Abstract

We present and study a mathematical framework to model cell to cell interactions in cancerous tissue. This frame work is called the Random Cell-cell Interaction Generator (RaCInG) and builds random graphs that represent cellular communication based on bulk RNA-seq data. In this talk we will first formalize the framework as a Monte-Carlo algorithm, and show that consistent biological properties can derived from it despite relying on some random choices to generate its output. Then, we analyze this framework mathematically and show it is asymptotically equivalent to a specific version of the inhomogeneous random digraph model. Thereafter, we will use this asymptotic equivalence to extract a large number of biological properties from a melanoma dataset and show that RaCInG is able to predict immunotherapy response within this dataset. Finally, we highlight that RaCInG is flexible enough to be generalized to other data modalities, and apply one such generalization to a glioblastoma cohort to showcase how it can learn spatially aware cellular communication patterns.

About Mike

Mike is a final-year PhD student at Eindhoven University of Technology. His research focusses on using random graph models to infer how cells in cancerous tissue communicate with each other. He is both interested in understanding these models on a fundamental level, as well as applying them to real-life datasets to understand the mechanisms in which tumors proliferate and evade the immune system. Apart from his research interests, Mike is a licensed high school teacher and is interested in developing new ways to teach and assess mathematics. During his PhD, he developed several new courses at the TU/e, and has researched the impact of alternative assessments to student engagement and performance.