Condensed Matter Physics Seminar: Probabilistic inference of immune repertoires diversity

Abstract:

The adaptive immune system can recognize many different pathogens by maintaining a large diversity of cells with different membrane receptors. We study the complex stochastic processes that generate and shape this ensemble of immune receptors developing probabilistic models from statistical inference of high throughput sequencing data. Our technique based on transfer matrices learns the probabilistic properties of the generation process, and finds it to be amazingly universal across individuals. We then model also selection pressures on the generated cells, in terms of the composition of their receptors, again finding universality, and reduction in diversity. In general our methods allows us to characterize and study the diversity distribution of immune repertoires using available sample data. This can be invaluable as a baseline for future study of the system as well as clinical applications, but might also expand our knowledge on statistical properties of interacting ensembles.

Seminar Organizer: Prof. Sasha Gerber