Computational Modelling Group

Seminar  25th November 2010 4 p.m.  University of Southampton, Building 59 (Zepler), Seminar Room 2

Chemotaxis and the sensing of molecular gradients as Bayesian inference

Duncan Mortimer
The University of Queensland, Australia

Web page
http://www.sense.ecs.soton.ac.uk/
Categories
Complex Systems
Submitter
Petrina Butler

Duncan Mortimer

http://www.physics.uq.edu.au/people/duncanm/Resume.htm

A SENSe Seminar: SENSe: Science and Engineering of Natural Systems at University of Southampton

Abstract

Molecular gradients are key sources of information for small biological systems. For instance, growing axons detect and respond to chemical cues to correctly wire up the brain, leukocytes follow trails of chemicals left by marauding bacteria to fight infections, and single celled organisms communicate and hunt by detecting nonhomogeneous molecular distributions in their environment. We frame the problem of sensing a gradient as one of Bayesian inference, and demonstrate that the variation of gradient sensing performance in several disparate biological systems is consistent with optimal estimation. Furthermore, we show that the polarisation of cells in a gradient --- a state in which cells tend to "turn" rather than abruptly change direction --- arises naturally in this framework as a consequence of the cell's expectations of the dynamics of its environment. We speculate that Bayesian computations might underpin behaviour even for individual cells.