Title:
Calibrating and validating the biological accuracy of a model neocortical column: from ion channels to network dynamics. Federation of European Neuroscience (poster 154.32 FENS). July 15, 2008, Geneva.

  by Markram H, Druckmann S, Gidon A, Hay E, King JG, Ramaswamy S, Ranjan R, Riachi I, Sfyrakis K, Schuermann F, Hill SL.

 Home  |  Back to Papers
 The Blue Brain Project has developed a simulation-based research environment for
modeling and studying the neocortical column (NCC) model of the rat somatosensory
cortex. A tool-chain has been constructed to build three-dimensional cellular
reconstructions of the neocortical column based on data from the somatosensory cortex
of the young Wistar rat. Models of ion channels, single neuron morphologies, electrical
firing properties and synapses are fitted to experimental measurements and used as the
building blocks of the column. The NCC model is composed of 10, 000 three-dimensional
reconstructed neurons arranged in mini-columns that extend throughout 6 layers of
cortex. Model ion channels are distributed on the model morphology to recreate
measured electrical properties of real neurons. Structural contacts between neurons
determine potential synaptic locations, while functional synapses are assigned - with
short-term synaptic properties - according the experimentally measured probabilities.
After the column is constructed, a calibration process for the neocortical column model
checks the biological fitness of: 1. layer boundaries, volume, density and composition of
the column; 2. single cell electrical behavior and ion channel kinetics; 3. dendritic
integration properties including backpropagating spike attenuation, linear/supralinear
summation and dendritic resonance; 4. morphology repair and cloning; 5. monosynaptic
properties including rise-time, amplitude and latency; 6. short-term synaptic facilitation
and depression; 7. polysynaptic loops including layer V pyramidal cell (L5PC) -Martinotti
and L5PC-L5PC interactions; 8. structural connectivity properties; and 9. emergent
phenomena including network oscillations and population responses to stimuli. This
modeling and calibration process highlights missing data, guides the acquisition of new
data, helps to define new experiments and improves the modeling process.
Posted on:17 Aug 2008