Evaluating automated parameter constraining procedures of neuron models by experimental and surrogate data

Abstract

Neuron models, in particular conductance-based compartmental models, often have numerous parameters that cannot be directly determined experimentally and must be constrained by an optimization procedure. A common practice in evaluating the utility of such procedures is using a previously developed model to generate surrogate data (e.g., traces of spikes following step current pulses) and then challenging the algorithm to recover the original parameters (e.g., the value of maximal ion channel conductances) that were used to generate the data. In this fashion, the success or failure of the model fitting procedure to find the original parameters can be easily determined. Here we show that some model fitting procedures that provide an excellent fit in the case of such model-to-model comparisons provide ill-balanced results when applied to experimental data. The main reason is that surrogate and experimental data test different aspects of the algorithm's function. When considering model-generated surrogate data, the algorithm is required to locate a perfect solution that is known to exist. In contrast, when considering experimental target data, there is no guarantee that a perfect solution is part of the search space. In this case, the optimization procedure must rank all imperfect approximations and ultimately select the best approximation. This aspect is not tested at all when considering surrogate data since at least one perfect solution is known to exist (the original parameters) making all approximations unnecessary. Furthermore, we demonstrate that distance functions based on extracting a set of features from the target data (such as time-to-firstspike, spike width, spike frequency, etc.)-rather than using the original data (e.g., the whole spike trace) as the target for fitting-are capable of finding imperfect solutions that are good approximations of the experimental data.

The use of automated parameter searches to improve ion channel kinetics for neural modeling

“…Both database and automated search methodologies have been used to explore the parameter space of neuron models (Keren et al 2005;Prinz et al 2003;Van Geit et al 2008). Due to the high dimensionality of a neuron model's parameter space, researchers often vary conductance density parameters while leaving channel kinetics fixed, which has been a successful method for matching many recorded features (Druckmann et al 2008;Gunay et al 2008;Prinz et al 2003). To extend this work, we set out to examine whether free channel kinetics can be constrained by automated searches and how much improvement is possible over density searches.…”

Efficient fitting of conductance-based model neurons from somatic current clamp

“…As discussed by Druckmann et al (2008), such robustness testing gives a far stronger validation than without perturbations, and has been proposed as a requirement for a method to 'hold promise to be successfully applied to experimental data'. The fitting method must then select a best estimate from a range of imperfect solutions, in contrast to the more commonly used reference model test where an exact solution always exists because the target and model to be fitted are of the same form.…”

Encoding the fine-structured mechanism of action potential dynamics with qualitative motifs

“…This could include adjusting parameters or functional forms of the gating variable equations to achieve an AP with different varieties of kinetics (e.g., without the use of a sodium inactivation variable). In contrast to related algorithmic approaches by Druckmann et al (2007), Druckmann et al (2008), Olypher and Calabrese (2007), Tien and Guckenheimer (2008), dominant scale analysis has the potential to adaptively determine salient features for optimization and reduce human supervision. model were: g L = 8.0, E L = −60, g CaF = 5, g CaS = 3.2, E Ca = 135, g K1 = 100, g K2 = 80, g K A = 80, E K = −70, g H = 4, E H = −21, g P = 7, E Na = 45, g Na = 200, I applied = 0, C = 0.5. transitions to be computed more accurately.…”