It is still not known how the worm Caenorhabditis elegans, a model organism in biology, coordinates its locomotion, even though the connectivity of its nervous system has been comprehensively mapped. One hypothesis is that the muscles are coordinated reactively according to the curvature of the body, through 'stretch receptors' on the sides of the worm. However, the suspected stretch receptors extend in the opposite direction to the direction intuitively required for locomotion—a paradox that casts doubt on this otherwise promising hypothesis.
Through computational modelling and simulation, I have shown that input from the suspected stretch receptors into the locomotion subcircuit is computationally sufficient for coordination of locomotion in a manner quantitatively similar to that of real C. elegans.
- Thesis (2.0 MB)
- Poster (480 KB)
- Simulation video (4.9 MB XviD)
An honours thesis submitted to The University of Queensland in October 2006. This is the most detailed documentation of this project.
A poster presented at the Queensland Brain Institute Workshop on Mathematical and Computational Neuroscience in August 2006. This provides an overview of the project. Many of the results and analyses in the thesis are not included.
A video of the simulation software simulating the locomotion of a C. elegans worm controlled by the stretch receptor-based neural network model.
SoftwareFor this project, a custom simulation, visualisation, and tools package called MuCoW (an acronym for Muscle Control of Worms) was developed. This software is available for download below.
In addition to implementations of the motor control and body models described above, this software includes:
- A fast Runge-Kutta physics integrator
- Modular architecture for runtime switching and configuration of motor control systems and body simulation
- An XML-configured parameter sweep mode, Headless MuCoW, allowing faster-than-realtime exploration of any region of the motor control or body model parameter space
- A parameter space visualisation tool, MuCoW Explorer, for displaying heatmaps of two-dimensional slices within multi-dimensional space (screenshot). Although this was designed for analysis of results generated using Headless MuCoW, it is based on a flexible CSV import system.
- An interactive realtime GUI (screenshot) for controlling and viewing simulations, including a highly customisable visualisation of the worm body and realtime graphing of metrics.
Boeing Prize for Best Software Project at the 2006 ITEE Innovation Expo.
To run, unpack and run the script in the root directory (mucow.sh or mucow.bat).
- Start the body physics simulation by clicking the button to the left of the worm body display. Reset the body at any time by clicking "Reset body (straight)".
- Use the drop-down list on the top-left to select a motor control system, then activate it using the nearby button. The "Programmatic central pattern generator" and "Neural network stretch receptor" models are discussed in the thesis. The other motor control systems are undocumented exploratory models only.
- Enable or tweak the body simulation visualisation by editing parameters in the "Visualisation parameters" tab to the left of the worm body display. Click on the options to see descriptions underneath the parameter list.
- Left-click-drag vertices on the worm body to interact with the body. Middle-click-drag up and down to zoom. Right-click-drag (when "Automatically centre view" is off) to move the view.
Note that MuCoW is not officially supported in any way and is made available for research purposes only. If you have any questions or comments, please email me.