Fitting experimental data to models that use morphological data from public databases

Area of Science:

• Neuroscience
• Computational Biology
• Biophysics

Background:

• Detailed neuron models ideally integrate morphological and electrophysiological data from the same cell, but this is rarely achieved.
• Current practices often involve using public morphology databases and standard parameter values, assuming model representativeness.

Purpose of the Study:

• To test the assumption that models with standard parameters and reconstructed morphologies yield representative experimental results.
• To investigate the impact of morphological variability and parameter fitting on neuron model accuracy.

Main Methods:

• Developed CA1 hippocampal pyramidal neuron models using four distinct morphologies from public databases.
• Employed the NEURON simulation environment's multiple run fitter to adjust parameter values against experimental data from 19 CA1 pyramidal cells.

Main Results:

• Models with fixed standard parameters failed to represent experimental data.
• Allowing parameter values to vary resulted in excellent fits, but fitted values differed significantly across reconstructions and deviated from standard values.
• Fitted parameter differences correlated with variations in cell diameter, length, membrane area, and volume, suggesting compensation for morphological discrepancies.

Conclusions:

• Neuron models incorporating reconstructed morphologies necessitate calibration with experimental data, even when data originates from the same cell.
• Generating model results using multiple reconstructions is crucial.
• Ensuring morphological and experimental cells originate from the same animal strain and age is important.
• Relying on standard parameter values without calibration may lead to unrepresentative model outcomes.