Capturing biological diversity: Modeling neuron and synapse heterogeneity in large-scale network simulations

Biological neural circuits are built from enormous diversity. Neurons differ in morphology, ion channel composition, and firing dynamics; synapses differ in strength, plasticity rules, and short-term dynamics; and these differences are distributed unevenly across cell types, layers, and brain regions. Classical large-scale network models have made substantial progress by abstracting away much of this heterogeneity, typically representing neurons and synapses as small numbers of canonical types. This has enabled simulations at impressive scale, but it leaves open a central question: how much does biological diversity itself contribute to circuit computation, dynamics, and function, and how can this be captured without making large-scale simulation computationally intractable?Recent progress in biophysically detailed modeling, data-driven parameter fitting, and high-performance simulation is beginning to close this gap. It is now possible to build networks that incorporate morphologically and electrophysiologically diverse neuron models, heterogeneous synaptic properties, and cell-type-specific connectivity rules, and to simulate these networks at scales approaching those of real circuits and regions. This progress raises new methodological challenges: how to efficiently simulate heterogeneous, biophysically detailed networks at scale; how to validate and benchmark these models against experimental data and against simpler, homogeneous alternatives; and how to build tools that make such simulations accessible and reproducible for the wider community.This Research Topic focuses specifically on the diversity of neurons and synapses as a modeling target in its own right, rather than as a complication to be abstracted away. We are interested in work that develops, applies, or benchmarks methods for simulating large-scale networks that retain biologically realistic heterogeneity, and in the tools, software, and standards that support this kind of modeling. We are also interested in how these principles of biological diversity extend into neuroscience-inspired artificial intelligence, offering an alternative lens on computation grounded in the brain’s own heterogeneity rather than in idealized, uniform units.We welcome contributions addressing, but not limited to:- Biophysically realistic modeling of diverse neuron types (morphology, ion channel composition, firing dynamics) within large-scale network simulations- Modeling of heterogeneous synaptic properties, including diverse plasticity rules and short-term dynamics, at network scale- Methods and algorithms for efficient simulation of heterogeneous, biophysically detailed networks, including parallelization, GPU acceleration, and multi-scale or hybrid simulation strategies- Data-driven approaches to constrain neuron and synapse diversity from experimental datasets (e.g., patch-clamp, single-cell transcriptomics, connectomics)- Neuroscience-inspired artificial intelligence (NeuroAI) and brain-inspired architectures that draw on biological diversity, including bio-inspired network topologies (recurrent connections, feedback loops, lateral interactions) and spiking neural networks with realistic temporal dynamics- Software tools, frameworks, and platforms that support building, simulating, or sharing diverse, large-scale neural network models- Benchmarking and validation strategies for comparing heterogeneous network models against experimental data, homogeneous model alternatives, or one another- Studies examining the functional or computational consequences of neuronal and synaptic diversity in large-scale circuit models- Standards, reproducibility practices, and open datasets/model repositories supporting heterogeneous large-scale simulation Topic editors: Arij Daou, Abhirup Bandyopadhyay, Diogo Leonai Souza, Jing-Xiao Liao Source: https://www.frontiersin.org/research-topics/85997/capturing-biological-diversity-modeling-neuron-and-synapse-heterogeneity-in-large-scale-network-simulations