|Title||Implementation of Cmicrocircuits model in NetPyNE and exploration of the effect of neuronal/synaptic loss on memory recall|
|Publication Type||Conference Paper|
|Year of Publication||2018|
|Authors||Tepper, Á., Sugi A., Lytton W. W., & Dura-Bernal S.|
|Conference Name||Computational Neuroscience Meeting (CNS 18')|
|Keywords||2018, BMC, BMC Neuroscience 2018, CNS|
The hippocampus has a major role in learning and memory, spatial navigation, emotional behavior and regulation of hypothalamic functions . Many models of its circuitry have been developed in order to further understand its functions . CA1 microcircuitry has been proposed to be responsible for the heteroassociative declarative memories  and the cycles of storage and recall are supposed to be modulated by theta oscillations  Cutsuridis et al.  modeled the CA1 microcircuitry using NEURON, the leading simulator in the neural multiscale modeling domain. The purpose was to investigate the biophysical mechanisms by which processes of storage and recall of spatio-temporal input patterns are achieved, employing a detailed biophysical representation of the CA1 microcircuitry. The model included five cell types whose functional roles were evaluated in the simulations. Each neuron had a specific morphology, ionic and synaptic properties, connectivity, and spatial distribution that closely followed experimental evidence. The original model was implemented in NEURON using HOC. The deprecated HOC language and the lack of standardization in NEURON makes it hard to understand, reproduce and manipulate and to run parallel simulations. Such a complex data-driven biologically realistic network would benefit from a separation of model parameters and implementation. To address these issues, we re-implemented the model using NetPyNE (www.netpyne.org), a high-level Python interface to the NEURON simulator, which facilitates the development, parallel simulation and analysis of biological neuronal networks . NetPyNE employs a standardized declarative format to describe the model specifications, and can then generate an efficiently parallelized NEURON model. It also provides a large number of analysis functions that enable further exploration of the model and allows exportation to NeuroML, a standard format for computational models. Our NetPyNE implementation is able to reproduce the results of the original model, but using a clean and powerful declarative language, which makes this complex model accessible to a wider community of neuroscientists. Furthermore, we analyse and explore the model in new ways, including connectivity analysis, computation of LFP spectra and information flow. We also perform novel manipulations to elucidate the relation between neuronal and synaptic loss, involved in Alzheimer's disease, and memory recall performance.