Combining sensitivity analysis with dimensional stacking to identify and visualize functional dependencies in conductance-based neuron model data
-
1
Humboldt-Universitaet zu Berlin, Institute for Theoretical Biology, Germany
-
2
Humboldt-Universitaet zu Berlin, Bernstein Center for Computational Neuroscience, Germany
-
3
Humboldt-Universitaet zu Berlin, Institute for Biology, Germany
Biophysically plausible neuron models are often characterized by a large number of parameters. Hence, visualization of observables derived from the simulation data, as a function of all underlying parameters is often difficult. However, visualization approaches exist, such as dimensional stacking (LeBlanc, 1990, Taylor et al., 2006), which maps a given discretely sampled (high-dimensional) parameter space onto a two-dimensional grid, that is amenable to visualization. This method does not omit information, as would, for example, a projection onto the first two principal components.
The usefulness of dimensional stacking for the identification of functional relationships depends on the choice of the so-called stack order. Only if the parameters with strongest influence on the observable are identified and ranked accordingly, the stacking method is suitable to extract information about the underlying system. Previous approaches (see, for example, Taylor et al., 2006) have solved this problem by sweeping through all possible stack orders and defining the optimal stack order as the one leading to lowest visual clutter in the resulting dimensional stack image. Here, we propose an efficient ranking procedure for the parameters by computing the impact of each parameter on the given observable. This allows a definition of the optimal stack order as the ranked impact scores, without the need of creating all possible dimensional stack images.
We demonstrate the usability of the proposed combination of sensitivity analysis and dimensional stacking, using the following example: we implemented a conductance-based single-neuron model with nine parameters that controlled the temperature dependence of the model's ion channel kinetics. Each parameter could take four values. We simulated the model at different temperatures and then computed the temperature dependence of, first, the stimulus-response curve and, second, of the energy consumption per action potential, for each possible parameter combination. Using sensitivity analysis, we show that the mechanisms underlying these two temperature dependencies rely on largely different sets of conductance parameters. This implies that both temperature-invariance and energy efficiency can be achieved in a single neuron without compromising each other. Dimensional stacking provides a direct visual estimate of the parameter impacts and also of the presence of higher-order effects.
Acknowledgements
BMBF 01GQ0901, 01GQ1001a, 01GQ0972, DFG SFB618, GRK 1589/1
References
LeBlanc J, Ward MO, and Wittels N. Exploring n-dimensional databases. (1990).
Proceedings of the First IEEE Conference on Visualization, edited by
Kaufman A, 1990, p. 230–237.
Taylor AL, Hickey TJ, Prinz AA, and Marder E. (2006). Structure and Visualization of High-Dimensional Conductance Spaces. J Neurophysiol 96:(2) 891-905
Keywords:
Conductance-based neuron modeling,
data visualization,
Energy-efficiency,
Grasshopper auditory system,
Temperature-compensation
Conference:
Bernstein Conference 2012, Munich, Germany, 12 Sep - 14 Sep, 2012.
Presentation Type:
Poster
Topic:
Data analysis, machine learning, neuroinformatics
Citation:
Roemschied
F,
Eberhard
MJ,
Schleimer
J,
Ronacher
B and
Schreiber
S
(2012). Combining sensitivity analysis with dimensional stacking to identify and visualize functional dependencies in conductance-based neuron model data.
Front. Comput. Neurosci.
Conference Abstract:
Bernstein Conference 2012.
doi: 10.3389/conf.fncom.2012.55.00138
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
11 May 2012;
Published Online:
12 Sep 2012.
*
Correspondence:
Mr. Frederic Roemschied, Humboldt-Universitaet zu Berlin, Institute for Theoretical Biology, Berlin, 10115, Germany, frederic.roemschied@bccn-berlin.de