Background Finding genes with a similar spatial expression pattern as a known gene could potentially reveal novel or unknown genes involved in similar processes or pathways. The Allen Brain Atlas (ABA) [1] is an effort to produce a genomewide mapping of the gene expression in the adult C57BL/ 6J mouse brain. To date, more than 21,000 genes have been assayed using a high-throughput in situ hybridization (ISH) platform and the resulting image data is publicly available at http://brain-map.org. A major goal of the ABA project is to employ image analysis techniques to search the ABA data for particular expression patterns such as spatial gene expression homologues. Methods Each ISH image series is processed through an automated anatomic mapping pipeline with the goal of determining expression sites and the spatial localization of these sites with respect to a 3D reference brain. Expression statistics is then aggregated with respect to individual 200 μm3 cubes in reference space thereby reducing data complexity from > 2 × 108 pixels per series to ~2.5 × 104 cubes. Since every image series is spatially mapped with accurate registration to the same 3D reference space, we can compare expression statistics on a global scale in approximately the same 200 μm3 spatial extent for all series. The most straightforward approach to finding spatial homologues from Sixteenth Annual Computational Neuroscience Meeting: CNS*2007 Toronto, Canada. 7–12 July 2007