Stable large-scale network architecture and structured individual variability in the awake mouse brain

Abstract

Large-scale functional networks provide a scaffold for brain activity, but their population architecture and inter-individual variability remain poorly defined in the awake mouse brain. Here we introduce the Awake Mouse Connectome Project (AMCP), a population-scale multimodal MRI resource comprising 212 sex-balanced awake mice. Using this dataset, we generated high-precision in vivo tissue probability maps, derived a reproducible whole-brain functional network atlas and quantified structured variability in network connectivity across animals. The resulting atlas defined a stable large-scale architecture aligned with structural topology, while inter-individual functional connectivity variability followed organized connectional gradients rather than random dispersion. We further benchmarked the clustering-derived scaffold against conventional group independent component analysis (GICA) and release the GICA network atlas as an accompanying resource. AMCP further provides empirical scaling relationships linking sample size, scan duration and connectome precision. These analyses indicate that mean functional connectivity and functional connectivity variability have distinct sample-size requirements, providing a protocol-specific guidance for studies focused on group-average connectivity and inter-individual variability. Individualized network parcellations derived from the AMCP scaffold improved behavioural prediction relative to group-average features for selected behavioural domains, indicating functional relevance of subject-specific network architecture. Together, AMCP establishes an open dataset, tissue-template resource, clustering-derived and GICA functional atlases, and analysis toolbox for reproducible awake mouse connectomics and for translational studies linking circuit mechanisms to large-scale brain organization.