Nanotode is a framework for running lightweight simulations of the C. elegans nematode's nervous system. It employs a compact binary representation of the neuronal connections of the organism, and is designed for use in resource-constained environments (e.g. microcontrollers).
Its spiking neural network utilizes a simplified leaky integrate-and-fire model, with inter-neuron connection weights derived from data available via the OpenWorm project [1].
It is my hope that this framework spurs further research (by myself and others) into how accurately we can render this tiny organism on a small computer—possibly inspiring a new set of educational tools for young people with an interest in biology or robotics.
-
CElegansNeuronTablesDirectory containing a CSV-converted representation of the OpenWorm project's CElegansNeuronTables.xls spreadsheet, which contains the layout of the C. elegans nervous system [2].
-
python_parsingContains utilities for parsing the aforementioned CSV files into a JSON representation, and parsing that JSON representation into a binary one suitable for consumption by the nanotode simulation (see file: source/neural_rom.c).
-
sourceHeaders and source files for the nanotode framework.
-
testContains an example main() function that ouputs a file representing the A and B-type motor neuron activity of the worm ('motor_ab.dat'), before and after sensory stimulation.
-
python_plottingTwo scripts for creating plots which visualize the data in a motor_ab.dat file
Figure 1.
A and B-type motor neuron activation in
the connectome simulation..
The source is configured to place the large array describing the nervous system found in 'neural_rom.c' into 'program memory' if it detects that it is being compiled for the Arduino UNO platform—otherwise, it is placed into normal RAM. For now, if you would like to compile for a different but similar Arduino platform, you will need to modify some code found near the top of the 'defines.h' file.
A simple WASM and SDL2 based game utilizing C. elegans simulations
An Arduino UNO-compatible robotic simulation of the C. elegans nematode
Busbice, T. Extending the C. Elegans Connectome to Robotics. URL: https://goo.gl/pxavvY
Busbice, T., Garrett, G., Churchill G. GoPiGo C. elegans Connectome Code. Github repository.
URL: https://github.com/Connectome/GoPiGo
Notes: Progenitor of this project.
-
Currently any cell that lacks post-synaptic connections is blindly categorized as a 'muscle' as per the source code's language (this is a purely cosmetic problem)
-
Only connections moderated by the GABA neurotransmitter are currently marked as being 'inhibitory' (i.e. having negative weight)—this could be more subtle
-
Neuron firing threshold and maximum idle time before falling back to zero 'potential' are essentially free parameters. What are the best ones? How do we define 'best'?
-
Is 'Number of Connections' (as per CElegansNeuronTables) the best first-order pass at inter-neuron weights? Would love to hear from an OpenWorm researcher about this.