2nd year results from the "Network Modelling" workpackage (WP 5)[]

The WP5 workpackage concerns the design, analysis and testing of models of neuronal networks. In this second year, we have continued work along the different tasks outlined in the proposal. The details are given in the progress report. We outline here the highlights of the results obtained collectively.

The FACETS project gathers a large number of different laboratories across Europe, and many of those groups are directly involved in designing simulations of network of neurons. This is one of the largest group ever put together with different modeling laboratories, and indubitably, this consortium is facing the problem of communication between so many groups. This is one of our main challenge in the modeling part.

To address this challenge, we have initiated two lines of action:

(1) Review paper on simulators. One of our biggest challenge concerns the fact that every group uses different types of models, and a different type of simulation environment. The first action was to write a detailed review of the different available simulators for simulating spiking neuronal networks. We have gathered the different partners of FACETS, and invited the principal investigators of the most prominent neuron simulation softwares, such as NEURON, GENESIS, XPP, NEST, etc. The article was accepted in The Journal of Computational Neuroscience. In addition to review and compare the available simulators, we provided in this paper a series of "benchmark" network models, implemented these models on every simulator, and provided the code of these models and made them freely available on the ModelDB database (http://senselab.med.yale.edu/modeldb/showmodel.asp?model=83319)

(2) A platform for programming models independent of any particular simulator. The PyNN interface was developed within FACETS, and enables us to write models using a language that is independent of the simulator. The user specifies which simulator is to be used. The same program code can thus be used for making the same simulation on different simulators. The web page for the PyNN project is at http://neuralensemble.org/PyNN .

In addition, we have progressed in network simulations to reproduce active states based on in vivo recordings, to simulate V1 responses, and to evaluate the network consequences of plasticity. This work is ongoing and will be reported in more detail in a later report.