Combining genome-scale Drosophila 3D neuroanatomical data by bridging template brains

The stereotyped structure of mammalian and invertebrate brains is a crucial determinant of their circuit organization. Thus large scale efforts to map circuit organization using 3D image data are underway in a number of model systems, including flies and mice. Many of these studies use registration of sample images to a standard template brain to enable co-visualization and spatial querying. However, studies often use distinct template brains, resulting in large islands of data which cannot be directly compared. To enable this comparison, we have constructed bridging registrations between template brains accounting for the vast majority of Drosophila melanogaster 3D neuroanatomical data. Furthermore, we solve the related problem of mapping data between the left and right brain hemispheres via the construction of mirroring registrations. Finally, we extend our approach across species to demonstrate its potential use in evolutionary studies of neural circuit structure and provide bridging registrations that link a new set of template brains generated for four Drosophila species that are divergent over 40 million years of evolution. We describe our strategy, document the freely available anatomical data and open source com- puter tools that we have generated and provide numerous examples of their use. This effort has unified data from over 30,000 publicly available images, with resources including the 3D atlas embodying the new standard Drosophila anatomical nomenclature and the largest single neuron databank yet available in any species. Over 20,000 registered images have been contributed to the Virtual Fly Brain project and can be viewed online at www.virtualflybrain.org.