Sarah Pallas / Sushil Prasad / Andrey Shilnikov / Igor Belykh
Title:
Modeling circuits for stimulus velocity tuning in the superior colliculus
Abstract:
A B&B grant would seed a collaboration between the investigators to explore the validity of a draft model of velocity tuning in visual neurons in superior colliculus (SC) of mammalian midbrain. Although several models of velocity have been suggested over the years, there has been no substantial improvement in matching possible models with the reality of the neurobiological data since Barlow and Levick's rabbit retina model in 1965. Because we have so much knowledge about the circuitry of SC (numbers of cells (Finlay & Pallas 1989), connectional anatomy (Pallas & Finlay 1991) and receptive field properties (Pallas & Finlay 1989) and about its developmental plasticity (Pallas & Finlay 1989, 1991; Xiong et al 1994, Huang & Pallas 2001; Razak et al 2003, 2005), we can provide more constraints on the parameters than is possible in other systems, e.g., in visual cortex. The models we produce for normal SC would be tested against what data is available from retina and cortex to determine whether they will generalize and to suggest modifications that provide a better fit in those brain regions.
Of greatest interest is to model how the SC compensates for perinatal damage or sensory deprivation. Tuning does not change under these conditions, and we have testable models of how this conservation occurs (Pallas & Finlay 1991; Huang et al 2005). The mechanisms for compensation in our model system are expected to generalize to recovery from brain trauma in general. Initial attempts at a computational model were carried out by Pallas group in collaboration with Norberto Grzywacz at USC (see Ascher and Grzywacz, 2000), using NeuroSIM. Due to the distance involved, that collaboration has foundered. The team proposes to capitalize on unpublished work already done by Pallas and Grzywacz and use this interesting and clinically-relevant biological problem to develop a computer-based model and efficient simulation programs. Typically, for simulation-based prediction, one carries out scores of simulation runs with various scenarios. In addition, a single simulation itself may run too slow if one scales to large/complex models. To address these, Prasad group would also explore the high-end computing based simulation techniques.
Andrey Shilnikov / Gennady Cymbalyuk
Title: Applications of the Poincare mapping technique to analysis of neronal dynamics
Abstract:
Understanding generic mechanisms of evolution and transitions between different patterns of activity of neurons is a fundamental problem for determining basic principles of the neuron functioning. We develop powerful suite of tools based on the Poincar return mapping technique to analyze both observable and
organizing hidden sides of dynamics of the membrane potential and a of slow channel currents in neural models. Knowledge of the structure of such a map allows one to study in detail periodic, aperiodic tonic spiking and bursting oscillations. One of the most critical problems in an experimental analysis of neuronal dynamics is that one can effectively measure only one state variable, the membrane potential, of all other ones determining the kinetics of ionic currents, i.e. activation and inactivation variables play crucial roles in
generating activity patterns. The method suggested here, which we will design and implement as an experimental procedure, should allow one to assess the dynamics of the slow variables by a particular analysis of the fast membrane potential V. We have received very exciting preliminary modeling data and want to expend it to the exposition level sufficient for an external grant application. The control parameter of the map has biophysical meaning, thereby allowing for accurate interpretation of the observable types of dynamics. Most excitingly promisingly this technique is directly applicable to the experimental in-line analysis.
Goal 1. To reconstruct numerically Poincare mappings for low-dimensional neural models. To analyze the mappings and to describe their dynamics and limit sets (attractors) under different conditions.
Goal 2. To conduct the systematic analysis, in terms of the Poincarè return mappings, of the onset and routes into bursting through tonic spiking, subthreshold oscillations and quiescence in the full model of the leech heart interneuron. This model demonstrates a variety of complex scenarios of such transitions, thereby indicating a global character of the underlying mechanisms and bifurcations describing the transitions. Transition from silence into bursting demonstrate features similar to that discovered in a simplified model but in addition demonstrate intermittent of the bursting activity – before neuron becomes silent it makes an undetermined number of bursts.
Goal 3. To create a database of the Poincarè maps for the dynamics in the neuronal models under different parameter conditions.
Goal 4. To record the return maps using the traces of living leech neurons in real-time electrophysiological experiments. To project experimentally recorded mapping onto those obtained in the models. This analysis will predict possible transitions between oscillatory regimes and reveal multistability of regimes if it is present.
Dr. Paul Katz / Rajshekhar Sunderraman / Ying Zhu
Title: NeuronBank: Knowledgebase of Identified Neurons and Synaptic Connections
Abstract:
The goal of the NeuronBank project is to create a flexible system for managing, analyzing, sharing, and extending our knowledge of identified neurons and the circuits they form. To accommodate the diversity of neural organization across species, we are developing the NeuronBank Species-Server Application. This application produces a web-based knowledge management system that can be easily customized to handle information about any type of nervous system. Researchers will be able to download this open-source application and immediately establish an online knowledgebase for storing and sharing information within their research community. To enable comparative analyses, we are also developing a NeuronBank Meta-Server, which will provide searches across published species databases. This federated knowledge-management system pushes forward several innovative computer science technologies and will serve an important need in the neuroscience community
|
