Microarray Data

The Data Set

The website offers three types of searches to allow a user to: (1) limit the visualization to a subset of probes (or genes) of interest (Gene Search); (2) use a 'seed' gene to find other genes with similar expression patterns (Find Correlates); and (3) compare expression between different anatomic regions (Differential Search).

Gene Search

To search for a gene category by disease, pathway, cell type or function, click on the relevant search term in the tag cloud. To search for probes associated with a specific gene or group of genes, select the Gene Search radio button and type the gene name, gene symbol or Entrez Gene ID in the search box (multiple genes searches require an OR between gene names) before clicking the Search button.

null

Boolean Syntax

The following special operators can be used to build queries:

• AND, OR and NOT may be used in place of their corresponding operators. They must be upper case.
• The AND operator (&) is implicit, so spaces between words that are not separated by an operator will be treated like an &.
• OR (|) has higher operator precedence than AND (&).
• Parenthesis can be used to group criteria, but nested parenthesis are not supported at this time.
• The NOT operator (!)is not supported within parenthesis.

Correlative Search

In using gene expression databases, a "search by example" feature is also highly desirable as genes with similar expression patterns may be related in function. The "Find Correlates" search utility will accomplish this function. This search by example facility is also available in the Allen Mouse Brain Atlas, Allen Developing Mouse Brain Atlas and BrainSpan: Atlas of the Developing Human Brain.

null

To find other genes with profiles similar to your gene of interest, first select your probe by clicking on any cell in the heat map related to that probe. You will see that probe listed in the box above the right hand side of the heat map. Then select the brain structure(s) in which you would like to see a similar expression pattern, and click "Find Correlates". This action will return probes with a similar expression profile to brain region(s) in which you are interested.

null

Only regions selected for the search will be displayed. To see the search results across the entire brain, turn off the "Restrict Domains" function at the bottom of the heat map.
You can see "anti-correlated" returns by toggling the sort order on column "r" or scrolling to the bottom of the heatmap.

Differential Search

Another common usage of gene expression databases is to find genes that show enrichment of expression in one region compared to another region. This type of query is supported by the "Differential Search" mode. Select the "Differential Search" radio button. To find genes or probes with a higher expression in one structure when compared to another structure, enter the target brain structure in the top search box and your contrast region in the bottom search box and click on the Search button.

Your search will return results in the two brain regions selected. The results will show higher expression in the target region compared to the contrast region. (To view the expression patterns of the returned probes over the entire brain, turn off the "Restrict Domains" function below the heat map.) Search results are sorted either by p-value or fold change, indicated by the arrow on the buttons over the column of genes. To alter the sort parameter, click on either the "p value" or the "fold-change" buttons.

To perform the previous search in reverse, click the toggle button beside the Search button.

Data Visualization

Based on your search, the resulting microarray data sets are presented as a matrix with brain structure (by individual donor) on the horizontal x-axis and gene probes on the vertical y-axis. The microarray data is presented in a "heat map" format where the colors of the heat map correspond either to raw data or to a normalized (z-score) expression level of a probe. Brain structures are organized such that moving left to right on the x-axis is analogous to moving from anterior to posterior first in the cortical areas, followed by subcortical areas, cerebellum and brainstem.

Data Aggregation and Normalization

Unless viewing data from a single sample, the heat maps presented on this site are based on data aggregated within brain structures ("Averaged Samples", see below). That is, when there are multiple samples for a given structure, the value represented in the heat map will be the average of those sample values.

We further aggregate the expression values up the ontology tree, i.e. Frontal Lobe, will have a single averaged expression value, which is the average value for all samples belonging to the Frontal Lobe.

Data represented in the heat maps has been normalized across the entire data set before it is aggregated, and is normalized again for each probe when the heat map is constructed.

Heat Maps

The "heat map" is a visualization of the microarray values for the returned probes of interest. Each row of the heat map represents a probe. Each column of the heat map either represents a tissue sample or anatomical brain structure depending on the selected Resolution (Coarse, Averaged Samples, or Sample). The colors of the heat map are normalized expression values. Default heat map colors are in the green – red scale where the color green should be interpreted as relatively low expression and red as relatively high expression within the scope of each probe.

The color map used for the heat map display can be changed to suit the user. You can use the color map bar at the bottom of the heat map to display different normalized (z-score) color representations of the heat map data (i.e. blue-red vs. green-red)or you can visualize log2 normalized expression map where the color scale ranges from dark blue, representing low expression, and passes through cyan, yellow, orange and finally to dark red, representing high expression.

Clicking on a cell of the heat map will bring up detailed information in the area above the heat map. In the panel on the left, information about the sampled anatomical structure is displayed. The stack of structures represents the hierarchy from the structure ontology. For more details on the ontology see Ontology and Nomenclature, or the Help section on the ontology viewer here.

Additional information is shown in the middle panel, and includes gene symbol, gene name, probe name, the "log2" expression value and z-score and links to related data in other Allen Institute projects. Links to donor meta-data are also included in this panel. Blue text indicates a hyperlink, where there is more information available by clicking on the text.
The "Brain Explorer" link, when selected for the first time, will take you to a page where you can download Brain Explorer® 2. Once Brain Explorer 2 is loaded, clicking on the Brain Explorer link will launch a desktop software application for viewing the Atlas gene expression data in three dimensions.

The "Planar View" link launches the multiplanar viewer that shows the expression profile for the selected probe in the context of the donor brain.

The columns of the heat map have a different meaning depending on the selected Resolution. There are three options available from the drop-down menu under the heat map: Coarse, Averaged Samples and Samples. If "Samples" is selected there is a one-to-one correspondence between a column and a physical tissue sample. In the Atlas, there are typically multiple samples for each structure of interest. This oversampling may provide information on variability and spatial gradients. In "Averaged Samples" mode, all samples belonging to the same designated structure are combined and averaged together. In "Coarse" mode, the brain is divided into approximately 20 large neuroanatomic divisions or regions (e.g. frontal lobe, occipital lobe, striatum, dorsal thalamus, ventral thalamus, etc.). Samples within each partition are averaged together to provide a summary value for the partition.

Data from multiple brains may be viewed in two ways. By default, columns are grouped first by donors then by structures, allowing a user to compare the expression profile of the individual brains side-by-side. Grouping the columns initially by structures then by donors allows a user to mix the data of the brains in the dataset within a combined brain structure profile. You can toggle between the two groupings by clicking the toggle button over the scroll bar.

Comparing Genes of Interest

You can add probes of interest to a collection for later viewing. Check the checkbox at the left end of a search result row to add it to your collection. Your choices are stored in a browser 'cookie' on your computer and will remain in effect until you click the 'Clear Selections' button, or clear your web browser's cookie cache.

Click the 'View Selections' button to see all of your selections.

Please note that this feature requires that you have cookies enabled in your browser, which is already the case for the great majority of users.

Gene Detail

Clicking on the Gene Name or Gene Symbol of your Search results will take you to the Microarray Gene Detail page.

The Microarray Gene Detail page displays information about the gene including Organism, Chromosome, Entrez Gene ID, related data from other Allen Brain Atlas resources, Gene Name and Alternate Symbols. Detailed expression heat maps for each probe are stacked by donor for a direct comparison between brains. There is also a section of links to External Resources.

Probe Detail

Clicking on the Probe Name will take you to the Microarray Probe Detail page.

The Microarray Probe Detail page displays information about the probe including Probe ID, Gene Symbol, Gene Name, Probe Name, NCBI Accession and GI numbers. It also includes the probe type, length, GC percent and sequence information. Detailed expression heat maps from each donor brain are stacked so as to directly compare the probe between brains. There is also a section of links to External Resources.

Multiplanar Viewer

The Multiplanar Viewer shows sampling sites for microarray data and indicates gene expression levels for a single probe in coronal, sagittal and horizontal sections of MRI space for a given specimen. Additional viewing frames show spatially correspondent histological data at the slab and block level with anatomic annotations and specific delineations of sampling sites for microarray data generation.

Multiplanar Navigation

In this image you see the Multiplanar navigation controls. Points from which microarray tissues samples were taken are colored to show expression level, according to the active color map.

Click or click-and-drag to move the crosshairs on any of the three MRI views ports.

The various control and feedback components are labeled here 1 - 7:

1. Donor: Identifier of the donor currently displayed in the viewer.
2. Probe Name: The label for the probe currently displayed in the viewer.
3. Brain Explorer Link & Permalink: Click the Brain Explorer link to go to the Brain Explorer view of this gene. Click the permalink component to generate a URL that will bring you back to this page with the current location settings.
4. Scale, MNI Coordinates & Contrast: Scale bar, MNI coordinates at the crosshairs. Click and drag to move the contrast control.
5. Brain Structure: This field is updated to indicate the structure under the crosshairs. Click on the structure label to launch the ontology browser in a new window.
6. Expression Level: This component reports both the raw expression value and its z-score-normalized value.
7. Color Map Window & Level: adjust the upper & lower bounds, or slide the entire control to modify the range of colors used to map expression levels.

Image and Probe Navigation

In this image you see the heat map associated with this probe as well as two image viewers illustrating the 2D images associated with the point selected by the MRI crosshairs. The viewer on the left hand shows a larger view of the brain specimen, with a blue highlight around the block of that specimen from which the current sample was taken. The viewer on the right hand shows a detailed view of that block, with structure annotations visible.

You can navigate from the image viewers as well. The strip of thumbnail images at the bottom of each viewer shows all of the images from its parent specimen, in sectioning order. Clicking on a thumbnail image in the left hand viewer will move the MRI crosshairs and open a new set of images in the right hand viewer. Click on one of the structures to move the MRI crosshairs to the location where that structure was sampled. As you move your mouse over the structures in the right hand viewer its name will appear in the upper-left corner of the viewer.

The toolbar icons provide more detailed views of the tissue sections: