Detailed biophysical neuronal models provide a unique window into the workings of individual neurons. They enable researchers to systematically and reversibly manipulate neuronal properties that are often impossible in real-world experiments.
These electronic models have played a key role in advancing our understanding of how neuronal morphology affects excitability and how specific ionic currents contribute to cell function. In addition, they play an important role in building neuronal circuits to simulate and study brain activity, giving us a glimpse into the complex dance of neurons that underpin our thoughts and actions.
Creating accurate electronic models that faithfully replicate experimental observations is no easy task. This requires quantifying the similarity between model responses and actual electrophysiological behavior, which is challenging when parameters such as ion channel conductance and passive membrane properties cannot be measured directly. Achieving high similarity scores often requires extensive exploration of the parameter space, which is arduous and time-consuming.
To address these challenges, researchers have turned to evolutionary algorithms (EAs) for help. Evolutionary algorithms are efficient tools for global parameter optimization in high-dimensional spaces. Among them, the indicator-based evolutionary algorithm (IBEA) has great potential in this regard. However, the field still lacks fully open source and replicable model optimization workflows.
In the new study, which appears on the cover of the November issue of Patterns, the BlueBrain Project proposes a groundbreaking common workflow for creating, validating, and generalizing detailed neuronal models. This approach is built on open source tools, with all steps freely available, providing researchers with a comprehensive solution for building neuronal models that can represent either individual biological cells or predefined cell types.
One of the unique features of this workflow is the ability to build so-called canonical neuron models. Werner Van Geit, leader of the BBP group, explains: "What we create is not a model customized for individual neurons, but a model that represents an entire neuron type. This approach is particularly useful when studying the properties of specific neuron types and building large neuronal circuits."
In this study, the authors applied a workflow to create 40 models representing 11 electrical types (e-types) in young mice's somatosensory cortex, the area of the cerebral cortex responsible for processing sensory information related to touch, pressure, temperature, and pain from various parts of the body. Each model is optimized based on a set of electrophysiological features to ensure a close match with experimental data. These typical models were then tested on various morphologies to assess their generality.
By analyzing the parameters used in these models, scientists gained insights into their biophysical properties. "Sensitivity analysis helps reveal which parameters are critical for model performance and which parameters can be varied without affecting the results," emphasizes co-first author Christian Rössert. "This deeper understanding goes a long way toward improving the creation of the model."
While this approach is powerful, the authors point out some current limitations. Certain neuron types generalize well across shapes, while others struggle. Understanding why certain models perform better in specific morphologies is an area of ongoing research. Furthermore, creating a single canonical model means ignoring some of the variability in real neurons. To solve this problem, neuroscientists can create multiple models based on the same inputs, introducing variation to represent real-world diversity.
Co-first author Maria Reva noted: "The set of electronic models presented here is based on electrical measurements from patch-clamp recordings of the main body of the neuron. In future versions, these models can be enriched with more details, such as synaptic and dendritic integration and additional ionic currents. These improvements will bring us closer to understanding the function of neurons."