The 2010 NEURON Simulator Meeting

News about new releases of NEURON, bug fixes, development threads, courses/conferences/workshops, meetings of the NEURON Users' Group etc.
Post Reply
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

The 2010 NEURON Simulator Meeting

Post by ted »

The 2010 NEURON Simulator Meeting will take place on March 22 and 23 at the University of Arizona in Tucson, AZ. It will be followed by a Course on Parallelizing Network Models with NEURON, which will start on March 24 and run through March 26. For detailed information about the Meeting and the Course, and an online registration form so you can sign up for one or both of these events, see http://www.neuron.yale.edu/neuron/stati ... m2010.html

Talks, symposia, tutorials, and workshops at the NEURON Simulator Meeting

The NEURON Simulator Meeting is very much a participatory event. All registrants are welcome to propose a talk, symposium, tutorial, or workshop (see instructions at http://www.neuron.yale.edu/neuron/stati ... m2010.html). The following posts list some of the presentations that will be given at the meeting.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

NEURON-RxD: Reaction-Diffusion Modeling in NEURON

Post by ted »

Title: NEURON-RxD: Reaction-Diffusion Modeling in NEURON
Format: Symposium
Speakers: Bill Lytton (organizer; SUNY Downstate), Juan Marco Alarcon (SUNY Downstate), Gwendal LeMasson (Bordeaux) Ion Moraru (University of Connecticut Health Center), Michael Hines (Yale University)

The focus of this symposium is NEURON-RxD, an extension of NEURON that is being developed in response to the growing interest in models of cells and networks that include mechanistic representations of key subcellular processes that underlie neuronal function. To meet this need, an extension of NEURON will be developed that can incorporate representations of reaction-diffusion processes at the genomic and proteomic level.

NEURON-RxD will incorporate much of the functionality of the VCell CSB modeling platform (http://vcell.org), and be able to simulate spatiotemporally complex biochemical interactions in cell bodies, dendrites and spines. Integration of the NEURON and VCell simulation environments will bridge between the computational neuroscience and systems biology communities, stimulating advances in our understanding of neuronal function across multiple scales. We will ensure interoperability with SBML (Systems Biology markup-language) and other markup languages as well as VCML.

The objectives of this symposium are to address the proposed design goals and feature set of NEURON-RxD, and to define specific research topics that can motivate and guide this work.

Introduction: Motivations and plans, and comparison to current simulators
Speaker: Bill Lytton

Title: Why we need whole-neuron (and whole-network) reaction diffusion modeling
Speaker: Juan Marcos Alarcon

Multiscale understanding of learning and memory in the brain will only be possible when we are able to integrate electrical activity at the cell and network scale with the complexity of the underlying synaptic and cellular mechanisms. This need is exemplified by the problem of Synaptic Tagging and Capture, a set of hypotheses that strive to explain how location and timing of inputs into the cell interact via internal chemical signaling to produce synaptic alterations at particular sites.

Title: Modeling axonal reaction-diffusion with NEURON, problems and potentials
Speaker: Gwendal LeMasson

Improved understanding and treatment of neurological diseases will require models that can interface neural activity with cascades leading to neural damage, some of which will be targets for therapy. In the case of Amyotrophic Lateral Sclerosis (ALS), there is evidence that axonal damage occurs due to a mismatch between metabolic demands for maintaining ion gradients and metabolic supply, which may be interrupted due to abnormalities in mitochondrial distribution. We will present a preliminary NEURON model that addresses this problem and describe some of the implementation problems that are encountered as we have begun to develop this model.

Title: Reaction-diffusion modeling: The VCell experience
Speaker: Ion Moraru

VCell is a cell simulator that has been in wide use in Computational Systems Biology for many years. Unlike several other simulators, the emphasis in VCell has always been on realistic simulation with diffusion, rather than assuming a "well-stirred" cell. We will consider some of the scientific and technical issues that arise in developing software of this type: modeling stochastic vs. deterministic diffusion, use of schematic vs realistic morphologies (compartmentalization strategies), intramembranous diffusion and registration at the membrane/cytosol interface, incorporation of active transport, etc.. We will also give a brief overview of the VCell user interfaces (GUI and programmatic) and comment on UI design decisions and advantages or limitations for particular project types.

Title: Bringing Rx-D to NEURON
Speaker: Michael Hines

Over the years NEURON has gradually expanded. This next step will require some reconceptualization of the simulator while of course retaining full backward compatibility. A key design aspect will be the need to maintain separation between geometrical and chemical kinetic specifications. The concept of sectioning/segmenting, which has been useful in the electrical domain, will be extended for compartmentalization for cytosolic diffusion. New front ends will also be adapted from VCell and elsewhere in order to facilitate model entry. The new simulator will be released in alpha stages during development so that users can begin to develop new models, and in doing so contribute suggestions as to additional needed features.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Parameter Optimization and Estimation

Post by ted »

Title: Parameter Optimization and Estimation
Format: Workshop
Speakers: Christina M. Weaver (organizer), Sean G. Carver, and potentially others

We will discuss current approaches to automated parameter optimization, and parameter estimation. Anyone having experience with these topics is encouraged to give a brief presentation of their work. Please contact Christina Weaver, (christina dot weaver at fandm dot edu) with your proposals.

Title: Parameter optimization of neuron models: lessons from the field
Speaker: Christina M. Weaver, Franklin & Marshall College, Lancaster PA

Mathematical models of neurons have many parameters, only loosely constrained within physiologically plausible ranges, that are difficult to estimate manually. Optimization algorithms simplify this task by automatically searching a multidimensional parameter space to identify combinations of parameters that best fit experimental data. These algorithms minimize a fitness function that represents salient differences between simulated and experimental data. Choosing an appropriate fitness function, and even parameter search method, is problem dependent. I will give an overview of NEURON's Multiple Run Fitter, and describe some fitness functions and search methods I have used to fit models to experimental data. Time permitting, I will also show how users may begin to design their own fitness function for NEURON, and how to use Matlab and NEURON together to aid the parameter search.

Title: Optimizing parameters with Bayesian Filtering
Speaker: Sean G. Carver, Ph.D., Yale University

We discuss a method for parameter estimation that can facilitate fitting a model to data in situations where it can be difficult to make any model follow the data. Such situations can arise, for example, when there noise in the biological system of interest that perturbs its state. Our method implements the well established technique in statistics known as maximum likelihood. We calculate likelihood with a type of Bayesian filter: an extended Kalman filter. These techniques are well known in other fields and starting to gain momentum in neuroscience. The method forces the model to approximately follow the data by adjusting the full state at each data point. This adjustment to unobserved states is done in a statistically principled way, based on the model, on the data, and on Bayes Rule. The amount of forcing required for the model to follow the data is quantified by the likelihood; this likelihood is the objective function for fitting. In this talk, we will explain what maximum likelihood is, how it is calculated, and why it can be a useful method for estimating parameters.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

A Practical Introduction to NEURON

Post by ted »

Title: A Practical Introduction to NEURON
Format: Tutorial
Speaker: Ted Carnevale

This course is designed for experimentalists and computational modelers who are interested in starting to use the NEURON simulation environment in neuroscience research. Through a combination of short talks, demonstrations, and hands-on exercises, it introduces the principles and procedures that are involved in creating and using models of biological neurons and networks with NEURON. Participants are invited to bring laptop PCs and Macs.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Failure to Communicate in the Brain

Post by ted »

Title: Computational Neuro-Matrimony: Dealing with Failures to Communicate
Format: Talk
Speaker: Jean-Marc Fellous, University of Arizona, Tucson, AZ

Individual neurons are very reliable devices, and cortical network computations can be extremely precise with spike timing in the order of 20 ms (e.g. phase precession, STDP). However, synapses in hippocampus and neocortex fail to transmit 3 times out of 4. Most conceptual and computational models do not account for this type of synaptic transmission variability. How can such noisy communication channels result in such reliable and precise neural computations? We present a new model of a stochastic synapse that includes facilitation and depression, and we show how the reliability and precision of spiking may be independently controlled by different types of synapses.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Hacking NEURON

Post by ted »

Title: Hacking NEURON
Format: Workshop
Moderators: Ted Carnevale and Bill Lytton

The aim of this workshop is to exchange "power programming" tips for getting the most out of NEURON. Discussion and specific examples will range over topics such as:
* useful idioms in hoc
* organizing programs for efficiency and clarity
* combining hoc code and the GUI to exploit the strengths of both
* hacking session files
* hacking NEURON's run-time system
* custom initializations
* automating execution of simulations and analysis of results
The moderators will present many of their own favorite hacks, and audience participation
is strongly encouraged.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Talk and tutorial on using NEURON with SAGE

Post by ted »

Title: NEURON simulations and Python prototyping using SAGE
Format: Talk
Speaker: Thomas McTavish, Yale University

SAGE is a mathematics software system that uses Python as its language and runs in a web browser on a user's local machine or over the internet. Since NEURON can be imported into SAGE, I will discuss how this tool can be used to prototype code, run NEURON simulations locally or over the internet, and display and analyze simulation output. I will also illustrate how this environment is useful for presenting, teaching, sharing, and collaborating.

Title: NEURON simulations and Python prototyping using SAGE
Format: Tutorial
Instructor: Thomas McTavish, Yale University

The objective of this hands-on tutorial is to familiarize participants with the basic features and possibilities of employing NEURON with SAGE. The tutorial will include worksheets that run basic NEURON simulations and analyze output, and do so over the internet. The worksheets are designed to demonstrate how SAGE can be useful for prototyping Python code, presenting data, teaching, sharing, and collaborating. Participants are encouraged to bring their own wifi-capable laptops.
Post Reply