Simulation Environments for Neuroscience Presented by Ali Nadalizadeh

ﺑﻪ ﻧﺎﻡ ﺧﺪﺍﻭﻧﺪ ﺑﺨﺸﻨﺪﻩ ﻣﻬﺮﺑﺎﻥ Simulation Environments for Neuroscience Presented by Ali Nadalizadeh – IPM – Summer 2009

Simulation Environments NEURON GENESIS NEST Neo. Cortical Simulator (NCS) Circuit Simulator (Csim) SPLIT XPPAUT (Discussed before)

NEURON - Introduction Historically, NEURON’s primary domain of application was in simulating empirically−based models of biological neurons with extended geometry and biophysical mechanisms that are spatially nonuniform and kinetically complex. (COBA Models) COBA – Stands for Complex Branched Anatomy

NEURON - Introduction COBA Models may include : Extracellular potential near the membrane Multiple channel types Inhomogeneous channel distribution Ionic accumulation and diffusion Second messengers NEURON can now simulate artificial models too, like integrate and fire or any combination of COBA and Artificial networks.

NEURON – How it works ? We'll approximate the continuous system of neuron into a discrete system in both time and space Every cell is constructed using connected Sections Every section is an unbranched, continuous cable whose anatomical and biophysical properties can vary continuously along its length Note that Sections differ from Compartments Neuron can do both Clock-Driven and Event Based simulations

NEURON - Other Features Different Integrate & Fire neuron choices Different Integration methods that will result in a tradeoff between speed and accuracy. Ability to define new biophysical mechanism. NMODL syntax → Compiled to C → Compiled to Native Machine code to be used by NEURON Runs under Windows, Mac. OSX, Linux Freely available plus extensive documentation

NEURON – Creating and using models Models should be written in an interpreted language named HOC NMODL language for new biophysic mechanisms A powerful GUI for conveniently building and using models Model. DB : Online model collection that are ready to use.

NEURON - Parallel Computing NEURON Supports 3 types of parallel computing Multiple Simulations on multiple processors Distributed Network models with gap junctions Distributed models of individual cells NEURON Uses MPI (Message Passing Interface) for parallel computing

GENESIS - Introduction Stands for General Neural Simulation System was given its name because it was designed, at the outset, be an extensible general simulation system for the realistic modeling of neural and biological systems (Bower and Beeman 1998) Typical GENESIS Models are multicompartment models with HH-type voltage/calcium dependent conductances Parallel Genesis (PGENESIS) is an extension for it, which supports MPI and PVM for parallel computing.

GENESIS – How it works ? Object-Oriented simulation system Message Passing Self-Knowledge (variables and actions) Neuron Component Examples Compartments Variable conductance ion channels Synaptic connections to other neurons

Model Neuron into compartments and compartments into circuits

Purkinje Cell Model With 4550 Compartments and 8021 channels

GENESIS - GUI

NEST Simulating neural networks of biologically realistic size and complexity Implementing a mathematically correct simulator by novices in a few days ? Reproducing the results of ad hoc simulations ! The NEST initiative was founded as a long term project to address these problems. Free/Open License with Extra conditions

NEST Easily copes with a threshold network size of 105 neurons with natural number of synapses No GUI (Network generation is usually procedural)

Neuro. ML Project A step toward making standard formats for neuron/network specifications. Current Standard Formats Morph. ML (specification of neuroanatomy) Channel. ML (specification of models of ion channels and receptors) Biophysics (specification of compartmental cell models, building on Morph. ML and Channel. ML) Network. ML (specification of cell positions and connections in a network. ) Common syntax of these specifications is XML

Python Integration What is python ? Py. NEURON, Py. Genesis, Py. NEST, Py. NN Benefits Standard Scripting Syntax Script Readability Neuro. ML Integration Ability to use current scientific tools (such as scipy)

References http: //www. open-mpi. org/ http: //en. wikipedia. org/wiki/Parallel_Virtual_Machine http: //www. neuron. yale. edu/neuron/ http: //www. wam-bamm. org/Tutorials/genesis-intro. html http: //www. python. org/doc/essays/blurb. html http: //en. wikipedia. org/wiki/Purkinje_cell Michael L. Hines, Andrew P Davison, Eilif Muller NEURON and Python Romain Brette · Michelle Rudolph · Ted Carnevale and other friends ! Simulation of networks of spiking neurons: A review of tools and strategies