Data-driven reconstruction of a point neuron mouse brain

In this work, we present a semi-automatic method to reconstruct a mouse whole-brain model at the point-neuron level by integrating a wide array of biological data. Our process has three parts: cell position and type assignment, connectivity mapping, and simulation. Additional validation is performed at every step of the workflow. We obtain cell positions from a voxelized data sets derived from high-resolution Nissl stained microscope image stacks in the Allen Mouse Brain reference atlas (Lein, et al. 2007), and global constraint numbers for the whole-brain (Herculano-Houzel, Ribeiro, et al. 2011). We then assign the type of each cell using In Situ Hybridization (ISH) image data from the same atlas. Cells are classified as glia subtypes, excitatory neurons, or inhibitory neurons, leaving the possibility to further expand the diversity of assigned cell types by using more genes in this step. We furthermore integrate region-specific cell densities from literature into our model. We then study cell type correlations throughout the brain and compare the resulting numbers to literature data in order to validate the process. In the second step, we use two-photon tomography images of recombinant Adeno-Associated Virus (rAAV) labeled axonal projections from the Allen Mouse Connectivity Atlas (Oh, et al. 2014) to determine the mesoscale connectivity between the neurons in different brain regions. For this step a comprehensive comparison to experimental data is difficult due to lack of similar connectivity data. We then obtain a network configuration that can be simulated with state-of the art software like Nest. We show results from a simulated whisker stimulation experiment and compare the evoked activity patterns to imaging data from Voltage Sensitive Dye (VSD) experiments (Ferezou, Haiss, et al. 2007). Finally, we build a glossary of possible whole-brain behaviors of our model, and briefly explore the possibility of a biologically plausible mapping of its input and output pathways.

Data-driven whole mouse brain modeling for multi-scale simulations

Dimitri Bruno Marie Rodarie

In this thesis, we present a data-driven iterative pipeline to generate, simulate and validate point-neuron models of the whole mouse brain. The ultimate goal is to replicate close loop experiments with a virtual body in a virtual world. This pipeline was origi-nally created by a PhD student of the Blue Brain Project and this pioneer work has yielded the first versions of the model and their simulations. My objective is to refine, extend and consolidate the previous version of the workflow and in particular to extend the repertoire of neuron types by integrating and consolidating available literature data.The pipeline has four main parts. First, we define a pair of reference atlases to create a spatial reference for every dataset that we want to integrate. Second, we create a cell atlas with estimates of the number and density of the major cell types in each brain region. Then, we build a connectivity atlas that describes the connections of each neuron of the mouse brain and their properties. Finally, we assign point-neuron parameters to each of the neuron types, and synaptic parameters to each of the connection types of our model to obtain a point-neuron network of the mouse brain. This network can be simulated to perform mouse brain experi-ments in-silico.In this thesis, we evaluate the quality of the different reference atlases, released by the Allen Institute for Brain Science and pro-vide methods to mitigate the impact of artifacts in the original images. We further extend the cell atlas with density estimates of more inhibitory neuron types. To do so, we introduce a new method to combine estimates of inhibitory neuron counts from the literature into a consistent framework. We also estimate inhibitory neuron density in regions where no literature data are avai-lable, using collected literature estimates and gene expression data from in situ hybridization image stacks. Our approach can be further extended to other cell types and provides a resource to build more detailed circuits of the mouse brain. Next, we refine the connectivity atlas, including literature findings on short-range connectivity. Additionally, we construct a database to store data collected on neuron types point-neuron parameters and synaptic parameters. With the results of the previous steps, we generate a new version of the whole mouse brain point-neuron network. We tested this new model against its previous version using three benchmark experiments. We explore the possibility to embed and connect detailed circuit reconstructions of the mouse brain to our point neuron model and co-simulate them. Finally, we constrain a 3D skeleton model of the mouse with joint constraints from the literature and attach muscle to its bones to create a musculoskeletal model of the mouse body that can be simulated in close-loop with our model to reproduce motor control experiments.

EPFL2022

Simulation Neurotechnologies for Advancing Brain Research: Parallelizing Large Networks in NEURON

Michael Lee Hines, Felix Schürmann

Large multiscale neuronal network simulations are of increasing value as more big data are gathered about brain wiring and organization under the auspices of a current major research initiative, such as Brain Research through Advancing Innovative Neurotechnologies. The development of these models requires new simulation technologies. We describe here the current use of the NEURON simulator with message passing interface (MPI) for simulation in the domain of moderately large networks on commonly available high-performance computers (HPCs). We discuss the basic layout of such simulations, including the methods of simulation setup, the run-time spike-passing paradigm, and postsimulation data storage and data management approaches. Using the Neuroscience Gateway, a portal for computational neuroscience that provides access to large HPCs, we benchmark simulations of neuronal networks of different sizes (500-100,000 cells), and using different numbers of nodes (1-256). We compare three types of networks, composed of either Izhikevich integrate-and-fire neurons (I&F), single-compartment Hodgkin-Huxley (HH) cells, or a hybrid network with half of each. Results show simulation run time increased approximately linearly with network size and decreased almost linearly with the number of nodes. Networks with I&F neurons were faster than HH networks, although differences were small since all tested cells were point neurons with a single compartment.

Mit Press2016

Axonal and Dendritic Morphology of Excitatory Neurons in Layer 2/3 Mouse Barrel Cortex Imaged Through Whole-Brain Two-Photon Tomography and Registered to a Digital Brain Atlas

Sylvain Crochet, Georgios Foustoukos, Yanqi Liu, Carl Petersen

Communication between cortical areas contributes importantly to sensory perception and cognition. On the millisecond time scale, information is signaled from one brain area to another by action potentials propagating across long-range axonal arborizations. Here, we develop and test methodology for imaging and annotating the brain-wide axonal arborizations of individual excitatory layer 2/3 neurons in mouse barrel cortex through single-cell electroporation and two-photon serial section tomography followed by registration to a digital brain atlas. Each neuron had an extensive local axon within the barrel cortex. In addition, individual neurons innervated subsets of secondary somatosensory cortex; primary somatosensory cortex for upper limb, trunk, and lower limb; primary and secondary motor cortex; visual and auditory cortical regions; dorsolateral striatum; and various fiber bundles. In the future, it will be important to assess if the diversity of axonal projections across individual layer 2/3 mouse barrel cortex neurons is accompanied by functional differences in their activity patterns.

FRONTIERS MEDIA SA2022

Data-driven whole mouse brain modeling for multi-scale simulations

Dimitri Bruno Marie Rodarie

EPFL2022