Abstracts

Parallel Computing Enables Full-Scale Modeling of the Control and Epileptic Rat Dentate Gyrus

Abstract number : 3.346
Submission category : 13. Neuropathology of Epilepsy
Year : 2010
Submission ID : 13358
Source : www.aesnet.org
Presentation date : 12/3/2010 12:00:00 AM
Published date : Dec 2, 2010, 06:00 AM

Authors :
Calvin Schneider, M. Case and I. Soltesz

Rationale: The dentate gyrus in the adult rat contains over one million neurons. Previously, we published a functional NEURON model of the rat dentate gyrus with over 50,000 biophysically realistic, multicompartmental neurons. With this model, we have studied the roles of network topological changes and highly interconnected hub cells in hyperexcitability. However, increases in synaptic conductances and connectivity were required to achieve realistic input to each cell in this 1:20 model. Enlargement of the model up to full size would eliminate these required scaling adjustments, and would represent a major advance in our quest to produce a biologically realistic, data-driven computational model of the rat dentate gyrus. Methods: To achieve the full-scale model, we rewrote our serial code to be compatible with the recently developed parallel NEURON simulation environment. The synaptic conductances and connectivity were adjusted to remove any scaling factors associated with the 1:20 network, and to allow for congruency with our previously determined structural model of the rat dentate gyrus. Additionally, the original length of the code was reduced by half and each line was documented to improve the accessibility of the model. Results: The parallel model was able to run on a variable number of processors and was modified to include a load balancing algorithm to ensure efficient use of computational resources. The full-scale model using parallel NEURON has been tested on two different clusters, UCI s Broadcom Distributed Unified Cluster (BDUC) and TeraGrid s Ranger Sun Constellation Cluster, providing a near linear speedup of computation time and dramatically increasing the overall memory available to the model. Conclusions: Through the modification of our model to be compatible with parallel NEURON, we have dramatically increased the computational power available to our model, providing us with the resources to model the full-scale control and epileptic rat dentate gyrus. This increased availability of computational power has also enabled us to incorporate more detail, to approach a more complete model of the dentate gyrus. Currently, we are including more cell types, adding gap junctions, and expanding the model from one to three dimensions. Through these improvements and model validation tests, we aim to construct a more complete full-scale model of the rat dentate gyrus, to provide a better tool to delineate the role of pathological changes observed in epilepsy on a functional and network level. This work was supported by the NIH (35915 to I.S.)
Neuropathology of Epilepsy