Skip to main content
Premium Trial:

Request an Annual Quote

Building a Better Brain Map


What was once an enigma is rapidly becoming charted territory. Thanks to classical histology, scientists have identified about 1,000 distinct structures in the mammalian brain — several hundred of which have been shown to have identifiable function. Many are a combination of small nuclei with variable cell types, particularly in the evolutionarily older regions of the brain such as the thalamus or medulla.


Modern neuroscience is utilizing methods of molecular genomics via gene expression and proteomics analysis to complement classical anatomy. This adds another layer of depth to understanding brain anatomy as, in principle, it becomes possible to map and classify anatomy down to the cell type level, an essential criterion in ultimately unraveling function.

Several powerful techniques have been developed to investigate gene expression in the mammalian brain, including microarray technology using laser capture micro-dissection, in situ hybridization (ISH) — both radioactive and colorimetric non-radioactive — and transgenic approaches using bacterial artificial chromosomes. Among these, ISH is perhaps unique in its ability to simultaneously offer cellular-level resolution and localization and the possibility of spatial anatomic mapping, as well as being suited to large-scale genomic style throughput.

Mapping hurdles

High-resolution spatial maps of gene expression can provide clues as to where genes regulate at neuronal and molecular levels. Localizing gene expression in vivo is less tractable, although there are some promising recent techniques under development. Much of neurogenomics currently focuses on in vitro work in the model organisms mouse and rat using 2D cryosections or by tissue dissection in which the spatial geometry is reconstructed via image registration and/or statistical averaging.

Central to the problem of mapping gene expression in the brain is the establishment of a common coordinate system that enables understanding complex spatial expression patterns in the same frame of reference. Performing queries such as “Show all genes expressed in the thalamus that are also expressed in the cerebellum” (see image) requires that the boundaries of these structures be identified with respect to a default standard brain where the anatomy is known a priori. Mapping can also place experiments from different modalities, times, and places onto an equal footing to enable cross comparisons. The standard way to approach this is through the construction of an anatomic reference atlas.

Classical neuroanatomy culminates in the modern atlases of Hof and Bloom, Paxinos and Franklin, and Swanson in which a stereotaxic coordinate system is employed to enable structure mapping and identification. These atlases for the mouse and rat brain have had tremendous value in providing the framework for researchers investigating normal, mutant, and transgenic animals. The atlases are typically two-dimensional, although the stereotaxic system enables comparisons across plane sections. To map genes in 3D, one adds a 3D reconstruction of the planar sections and a digital reference atlas. Digital atlases are preferred because they allow arbitrary virtual re-sectioning and 3D visualization.

Constructing 3D atlases can be approached in a variety of ways that include hand and automated annotation, and by combining multi-modality data. The most advanced atlases are presently for the mouse and rat using techniques as diverse as MRI, PET, in situ hybridization and immunohistochemistry. Once the atlas is constructed, some form of image registration or deformation algorithm is used to transfer the atlas annotation back onto the primary data. This key step solves the structural mapping problem and enables researchers to query measurable quantities that summarize expression in each anatomic region.

By mapping the brain it becomes possible to enter quantitative information into a database and to pose both textual and spatial image-based queries. This allows searching for common patterns of gene expression and is a clear advance over string- and text-based methods. Presently, most of existing gene expression data is known only in textual terms by reference to anatomic ontology, and in general anatomical descriptions will be limited in detail and less specific than image data. While considerably more difficult to effectively implement, spatial searchable image databases overcome these limitations.

The Allen Brain Atlas project has surveyed the C57Bl/J6 mouse genome for expression patterns and mapped in 3D more than 20,000 genes into a searchable online reference atlas. This required the development of a high-throughput pipeline to achieve cross modality ISH to Nissl image registration without reliance on user-added anatomic landmarks and with the processing capacity of 1 terabyte per day. A primary goal of the pipeline was to acquire co-registered data on the expression of individual genes in a fashion that enables online anatomic structural search, visualization, and data mining. Specific challenges in this mapping led to developing modules that support:

• Image preprocessing, including 10x magnification tile stitching and direct compression into JPEG2000 format. The inherent multi-resolution format of JPEG2000 as well as its robust wavelet-based compression makes it an ideal choice for image storage and transport.

Generation of a novel 3D online digital reference atlas for the adult C56Bl/6J mouse brain. A 3D virtual atlas is lofted from 2D hand annotated anatomic drawings.

Fully automated 3D image reconstruction and deformable registration to bring the ISH images into a common anatomic framework with Nissl stained images.

• Signal segmentation and estimation for quantification of expressing cells and tissues in ISH images. Results are summarized in heat maps of expressing tissue for each ISH section.

•Visualization tools for examining 3D expression patterns of multiple genes in anatomic regions. (Download and try the public application Brain Explorer, a tool developed to search, browse, and analyze expression patterns, at

Mike Hawrylycz is Director of Informatics at the Allen Institute for Brain Science. His group is respon-sible for the neuro-informatics of atlas mapping and data analysis.


Due to the increasingly central role of imaging in neuro-genomics research, major advances are being made in the areas of data organization, mapping, and quantification of imaging data. Mapping gene expression in the brain on a large scale challenges each of these areas. To learn more about this exciting and rapidly advancing area of neuroinformatics, check out the following sites:

• Edinburgh Mouse Atlas Project,

• Geneatlas,

• Laboratory for Bioimaging and Anatomical Informatics

• Mouse Atlas Project,

• Allen Institute for Brain Science,


The Scan

Genetic Risk Factors for Hypertension Can Help Identify Those at Risk for Cardiovascular Disease

Genetically predicted high blood pressure risk is also associated with increased cardiovascular disease risk, a new JAMA Cardiology study says.

Circulating Tumor DNA Linked to Post-Treatment Relapse in Breast Cancer

Post-treatment detection of circulating tumor DNA may identify breast cancer patients who are more likely to relapse, a new JCO Precision Oncology study finds.

Genetics Influence Level of Depression Tied to Trauma Exposure, Study Finds

Researchers examine the interplay of trauma, genetics, and major depressive disorder in JAMA Psychiatry.

UCLA Team Reports Cost-Effective Liquid Biopsy Approach for Cancer Detection

The researchers report in Nature Communications that their liquid biopsy approach has high specificity in detecting all- and early-stage cancers.