Scientists have achieved a major breakthrough in computational neuroscience by creating one of the most detailed and biologically realistic simulations of the mouse cortex ever developed. Using a biophysically rich, neuron-by-neuron digital reconstruction, the model captures sub-cellular activity, including ion flows and membrane voltage fluctuations across the many compartments of real neuronal morphologies. Even though only 1% of the neurons are visualized for clarity, the system simulates spontaneous resting-state activity with lifelike fidelity. Each virtual neuron is color-coded by cortical region and flashes during activation, offering a striking representation of how real neural circuits operate across the entire mouse cortex.
This achievement was made possible by Supercomputer Fugaku, one of the world’s fastest high-performance computing systems. Fugaku’s immense processing capacity—exceeding 400 quadrillion calculations per second—enabled the simulation of almost ten million neurons interconnected by 26 billion synapses and organized into 86 brain regions. Such computational power is essential for recreating the structural and functional intricacies of the mouse cortex at this unprecedented resolution. The project is a collaboration between the Allen Institute, Tadashi Yamazaki’s team at Japan’s University of Electro-Communications, and several Japanese research partners. The full technical report will be presented at SC25, the world’s premier supercomputing conference.
Beyond its technical sophistication, this virtual cortex offers a powerful new tool for studying brain disorders and cognitive processes. Researchers can now explore how neurodegenerative diseases like Alzheimer’s emerge, how seizures propagate through neural networks, or how brain-wave dynamics relate to attention. Previously, such investigations required physical tissue and invasive experiments. Now, scientists can test hypotheses in a controlled digital environment, accelerate discovery, and assess therapeutic strategies before moving to biological studies. According to Allen Institute investigator Anton Arkhipov, this milestone proves that massively scaled brain simulations are not only feasible but ready to expand even further.
Constructing the digital mouse cortex required integrating vast biological datasets with advanced computational tools. The Allen Institute supplied detailed information from the Cell Types Database and Connectivity Atlas, forming the biological blueprint for the model. On the computational side, Fugaku’s 158,976 interconnected nodes handled the immense workload, while the Brain Modeling ToolKit and Neulite software converted mathematical neuron equations into simulated cells capable of signaling, spiking, and interacting like their real counterparts.
The resulting simulation mirrors living brain activity with remarkable accuracy. Viewers can observe neuronal structure, synaptic behavior, and electrical signaling across membranes, capturing the complex choreography of cortical communication. Yamazaki emphasizes that although this is a remarkable technical accomplishment, it represents only the beginning: attention to biological detail will remain crucial for increasingly realistic models.
Looking forward, the team aims to construct whole-brain simulations, eventually scaling to human models. With continued advances in computational power and biological data collection, fully comprehensive and biophysically accurate brain reconstructions may soon become a reality. This work signals a new era in neuroscience—one in which understanding the brain includes the ability to build one from the ground up.
https://www.sciencedaily.com/releases/2025/11/251118212037.htm