Dr Gabriel Rosser

My research interests concern the motile behaviour of freely-swimming bacteria.  Most bacterial species can swim through a liquid medium, mediated by rotating semi-rigid filaments known as flagella.  Because they are so small, bacteria sense the environment temporally not spatially, and use a biased random swimming pattern through their viscous environment to reach optimum environments [3].  Different species have evolved different patterns of swimming, and different flagella numbers, from one to over 30 per cell.  Bacterial motility is a key factor in surface colonisation and the formation of bacterial biofilms.  Biofilms are of critical clinical and industrial importance; they are a major cause of infection and contamination in hospitals [1], and cost US oil industries billions of dollars each year by clogging pipelines [2].

One of the most common protocols for studying bacterial motility involves observing bacteria swimming under a microscope and tracking their movements [4].  The tracks obtained contain important information on the swimming patterns of that species and the variation in motile behaviour both within and between individual cells.  The current state of the art allows us to obtain a large number of tracks with relative ease and low cost. However, the data must be analysed appropriately in order to extract reliable statistics relating to the observed motility.

Analysis methods for bacterial tracking data

My experimental collaborators have developed an inexpensive and high-throughput protocol to track free-swimming bacteria. I develop novel methods for characterising the observed swimming patterns.  For example, bacteria such as Rhodobacter sphaeroides exhibit random reorientation events, accompanied by a short stationary phase, separated by approximately straight swimming phases.  The stopping phases are hard to determine automatically, as various sources of noise are present in the data.  I have implemented a state space approach to identify these phases in tracks.

Mathematical models of bacterial motility

A key question in the study of bacterial motility is the role of intrinsic and extrinsic noise. Using a velocity jump modelling framework [5], I am undertaking a theoretical study of the effect of Brownian buffeting and measurement noise on the observed process.  These two models of noise differ fundamentally in their effect on the recorded tracks.  Measurement noise is applied independently of the true underlying motion, whereas buffeting affects this directly.

Modelling-based optimisation of experimental protocols

In addition to providing insight into biological behaviour, mathematical models enable us to quantitatively study the effect of making changes to the experimental protocol. I have used this approach to investigate the way that the sampling frequency of the capture device alters the nature of the tracks obtained.  An overly low sampling frequency means that the obtained tracks lack detail.  Conversely, a very high sampling frequency may lead to tracks dominated by measurement noise.

[1] J. W. Costerton, Philip S. Stewart, and E. P. Greenberg. Bacterial biofilms: a common cause of persistent infections. Science, 284:1318–1322, 1999.
[2] Bernard Ollivier and Michel Magot. Petroleum Microbiology. Amer Society for Microbiology, 2005.
[3] Howard C. Berg. Random walks in biology. Princeton University Press, 1993.
[4] P. Poole, D. R. Sinclair, and J. P. Armitage. Real time computer tracking of free-swimming and tethered rotating cells. Anal. Biochem., 175:52–58, 1988.
[5] H. G. Othmer, S. R. Dunbar, and W. Alt. Models of dispersal in biological systems. J. Math. Biol., 26:263-298, 1988.