Mouse Connectivity Models (MCModels) is a Python package providing mesoscale connectivity models for mouse using data from anterograde viral tracing experiments. This voxel-scale model of the mesoscale structural connectivity in mouse permits researchers to extend their previous analyses to unprecedented levels of resolution and allow for comparison with functional imaging and other datasets. This package was initially developed for a voxel-scale connectivity model described in the publication below and has grown to include multiple resources for exploring and modeling brain connectivity.
In the publication "High resolution data-driven model of the mouse connectome," a method is presented for modeling mesoscale brain-wide connectivity at the resolution of voxels. Using wildtype rAAV anterograde viral tracing data registered to a 3D brain volume, a spatial interpolation method was used to infer the connectivity from every voxel in the mouse brain. This image shows the modeled cortical connectivity from a source (red dot), mapped to the surface of the cortex.
This voxel-scale model can also be used to study connectivity at coarser levels of resolution, such as regional connectivity. The anatomical parcellation from the Allen Brain Reference Atlas for adult mouse allows integration at the voxel-level, to determine connectivity weights within each region. In the publication cited above, a regionalized connectivity matrix was used to reveal community structure of the mouse cortical network, where the model provided information to predict projectional pathways for regions lacking well-isolated injection experiments. More detailed information and code associated with this model is available through the MCModels site on Github.
The voxel-scale model is a connectivity matrix with approximately ten billion entries. This model inherently produces a low-rank representation of this very large matrix, allowing researchers to build voxel-scale connectivity models for smaller sections of the brain. Building the connectivity matrix is possible through the MCModels package. This process is outlined in a user guide and detailed description in the API available here.