Links
The idea and endeavour of simulating C. elegans is not new. It has been suggested several times (e.g., NemaSys (US, ~1997), Virtual C. elegans Project (Japan, ~2004)) and is currently pursued by the openworm project (US). While most models are restricted to neurophysiological aspects, the closed-loop interaction between neural circuits and biomechanics has been recently addressed as well. Beside the utmost detailed and faithful geometrical virtual model of Caenorhabditis elegans (with all of its cells) published by Grove and Sternberg (Grove & Sternberg, 2011)1, currently two biological accurate 3D models of C. elegans locomotion are available (Boyle, 2009; Mailler et al., 2010). Current research aims at the constant refinement of the body simulation through manipulation experiments of alive C. elegans nematodes and their extremely fine video tracking and processing to validate simulation accuracy. Towards a holistic approach, where the behaviour of an animal is modeled through the interaction between neurons, muscles and the environment, Voegtlin et al. have developed software that can simulate the interaction between neurons and muscles efficiently. It is based on an open-source library called CLONES (Closed Loop Neural Simulation), a communication interface between the BRIAN neural simulator (Goodman & Brette, 2008) and SOFA, a physics engine for biomedical applications (Allard et al., 2007). Most of these activities are integrated in a more general context, the OpenWorm project. It aims at providing a full simulation of the nematode with a focus on capturing as much of the rich detail of its biological system as possible. This project is based on results obtained in other major projects, namely neuroConstruct, the Simulation Framework, Genetic Algorithms and Physics. In neuroConstruct, all 302 neurons of C. elegans described in the virtual worm have been described as multi-compartmental neuronal models using NeuroML descriptors (computational models based on detailed neuroanatomical and electrophysiological data). The Simulation Framework is a multi-algorithm, multi-scale simulation platform engineered to support the simulation of complex biological systems and their environment. Each of the 302 neurons is simulated in parallel on GPU native kernels via OpenCL. The simulation engine can be distributed across multiple machines.
Modelling neural systems has diverse roots and inspirations, most of them being inductively derived from first principles (e.g., Hodgkin & Huxley) or deduced from direct observation. Several proprietary (e.g., MATLAB) and open-access neural simulation software platforms have been created over the years, including NEURON, GENESIS , MOOSE, NEST, NeuroML, Python, NNetWARE, BRIAN, PSICS and some modelling environments designed particularly for execution on parallel computer architectures (e.g., namely SpiNDeK (Hauptvogel et al., 2009)), to name a few prominent ones. These modelling tools allow the non-real time simulation of individual neurons or small network assemblies on single-processor or multiprocessor machines. The principal approaches used are compartmental models for 'ab-initio' simulations or more integrative strategies based on integrate-and-fire spiking models with add-on modalities such as leakage, decay or delays.
Wrapper programs have been developed (including open-access initiatives such as NEST, PyNN, ConnPlotter (Nordlie & Plesser, 2010), NineML ...) towards ease of handling and standardization of model generation procedures.
Recent endeavours with an eye on in-silico implementations of neural functionality include EU-funded projects such as PERPLEXUS, FACETS, CAVIAR (Serrano-Gotarredona et al., 2009), SpikeFORCE, REALNET, Brain-i-Nets, BrainScaleS, POEtic, eMorph, NeuroP, Sensemaker, Blue Brain including the recent Human Brain Project EU flagship initiative and national endeavours like Janus (Spain/Portugal), SpiNNaker (UK), neuFlow (US), EPSRC Spiking Neuron Project (UK), SyNAPSE (US), Neurogrid (US), HP's Cog ex Machina (US), MoNETA, IBM's C2 (US), IFAT 4G (US), and statements of intent such as Grand Challenges for Computing Research (Sleep, 2003). Current and past initiatives and trends with a discussion of some of the reasons why the brain emulation hasn't progressed beyond the state of the art are summarized in (De Kamps, 2012).
Surprisingly, despite these efforts, there is no emulation that approximately, sufficiently and flawlessly replicates and mimics a simple biological central nervous system like that of C. elegans.
1 The morphology, arrangement and connectivity of each cell including neurons have been completely described and are found to be almost invariant across different individuals. All of these data have been published and are freely available, e.g., through www.wormatlas.org. A complete digital representation of the anatomy of C. elegans can be found @ caltech.wormbase.org/virtualworm. Also see Mailler et al., 2010