The gene expression database for mouse development (GXD): putting developmental expression information at your fingertips

Abstract

Because molecular mechanisms of development are extraordinarily complex, the understanding of these processes requires the integration of pertinent research data. Using the Gene Expression Database for Mouse Development (GXD) as an example, we illustrate the progress made toward this goal, and discuss relevant issues that apply to developmental databases and developmental research in general. Since its first release in 1998, GXD has served the scientific community by integrating multiple types of expression data from publications and electronic submissions and by making these data freely and widely available. Focusing on endogenous gene expression in wild-type and mutant mice and covering data from RNA in situ hybridization, in situ reporter (knock-in), immunohistochemistry, reverse transcriptase-polymerase chain reaction, Northern blot, and Western blot experiments, the database has grown tremendously over the years in terms of data content and search utilities. Currently, GXD includes over 1.4 million annotated expression results and over 260,000 images. All these data and images are readily accessible to many types of database searches. Here we describe the data and search tools of GXD; explain how to use the database most effectively; discuss how we acquire, curate, and integrate developmental expression information; and describe how the research community can help in this process.

Keywords: anatomy ontology; data integration; immunohistochemistry; in situ hybridization; literature curation; online resource.

Figure 1

Left: GXD assay details pages provide detailed annotations of experimental parameters and expression results. Shown is an example of an Assay Detail page for an immunohistochemistry experiment illustrating the detailed content of expression results in GXD. The Assay section reports the reference from which the data were derived, the assay type, the gene analyzed, and the antibody used, with links to more details about the antibody, reference, and gene. The Results section reports the tissue (Theiler stage and anatomical structure) analyzed, as well as the level and pattern of expression, as described by the authors. Images of the original expression data are displayed together with the corresponding annotations whenever possible. Major specimen details such as the age and mutant alleles are always displayed on this page. Other information, such as genetic background, sex, and specimen preparation method, is accessible via the ‘more’ toggle. Right: Image detail pages are accessed by clicking on the image panes or pane labels shown in the assay details record. They show the entire figure, as published, providing the scientific context. As shown in the Associated Assays table of this example, several genes were studied in the same or similar tissue sections; the assay IDs and gene symbols link to the corresponding annotations for each image pane. While the image detail pages provide visual context, it is the detailed and standardized text annotations shown on the GXD assay detail pages (left) that make the expression data, including image data, accessible to searching.

Figure 2

Gene Detail pages summarize, and provide access to, all the information about a given gene in MGI, together with extensive links to external resources. The upper portion of the Otx2 gene detail page is shown. The expression section (expanded) indicates the types and amount of expression information available for the gene and provides links to the corresponding summary pages. Links to databases that store mouse expression data not available in GXD are provided as well: the Allen Institute (Lein et al., 2007), GENSAT (Heintz, 2004), GEO (Barret at al., 2013) and ArrayExpress (Petryszak et al., 2014).

Figure 3

The Gene Expression Data Query Form features two search tabs: Standard and Differential Expression. The Standard Search, shown here, enables queries for expression data using one or more parameters. The Genes section allows users to find expression data for a specific gene or for a set of genes based on their function [as defined by Gene Ontology terms (Gene Ontology Consortium, 2010)], their association with mouse phenotypes [as defined by Mammalian Phenotype Ontology terms (Smith et al., 2012)], or their association with human diseases [as defined by Online Mendelian Inheritance in Man (OMIM) terms (Amberger et al., 2011)]. In the anatomical/stage section, one can search for expression data in specific anatomical structures and/or developmental stages, and one can specify whether (1) all results should be returned or only those where expression was (2) detected (i.e. present) or (3) not detected (i.e. absent). Anatomical searches combine word searching and hierarchical searching. For example, a search for expression in “diencephalon” would return expression annotations for all anatomical structures that have “diencephalon” as part of their name as well as for all their anatomical substructures such as “thalamus”. In the mutant/wild type section one can limit the searches to expression data from wild-type mice or search for gene expression in specific mutants. The Assay types section allows selection of expression data types. Auto-fill utilities help to find appropriate search terms. The illustrated search asks for ‘transcription factor binding’ genes ‘detected’ in the ‘diencephalon’ at ‘Theiler stages 17, 18, or 19’. The corresponding search results page is shown in Fig. 4. The Differential Expression Search (not shown) allows searching for genes that are expressed in some anatomical structures but not others and/or at some developmental stages but not others.

Figure 4

GXD data summaries can be viewed at different levels of detail and interactively refined and sorted. Searches using the Gene Expression Data Query form return a page with four tabbed summaries for the assay results, assays, genes, and images that match the search parameters. The assay results tab (upper) is displayed by default. It lists the gene studied, the assay type used, the anatomical system, age and tissue examined, indicates whether expression was detected, provides a link to the corresponding images, lists the mutant alleles of the specimen (if applicable), and provides the reference from which the data were derived. Links in the Result Details and Images columns lead to detailed expression records, such as the one shown in Fig. 1. Arrows in column headers indicate that the column is sortable (one set is circled). The assay results tab (as well as the genes tab) allows for the export of results in text and spread sheet formats (buttons in table header). The images tab (lower) shows all the images that match the search criteria, together with the gene(s) examined in that image and the assay type used and provides a link to the corresponding part of the detailed expression record. The expression summaries can be refined by using the ‘click to modify search’ button or by employing the filter options provided on the summary page. The content of all four tabbed summaries will change accordingly.

Figure 5

The Mouse Developmental Anatomy Browser allows users to search for anatomical terms, to explore the anatomical hierarchies and locate specific anatomical structures in context, and to retrieve the expression data associated with these structures and their substructures. The anatomy search is facilitated by an auto-fill utility. As soon as a term is selected from the pick list, all matching anatomical structures are displayed in the search column, together with the developmental stage range during which these structures are present in the embryo. The best match is listed first and selected by default. Other matching terms can be selected by clicking. Upon selection, the Anatomical Tree View and the Anatomical Term Detail section are updated and the selected anatomical structure is highlighted. Using the Tree View, users can explore the ontology further by expanding and collapsing branches. Clicking on a term in the tree view will select (and highlight) that term. The number of expression results associated with each term is listed; following that link will lead to an expression summary page similar to the one shown in Fig. 4. The initial tree view shows the abstract version of the anatomy ontology. Accordingly, the associated expression results will include the annotations for all developmental stages at which the selected anatomical structure is present. The developmental stage pick list in the Anatomical Term Detail section allows users to toggle between stage-independent terms and tree views and stage-specific terms and tree views. Stage-specific terms will link to the expression results for the anatomical structure at that specific stage.

Figure 6

Querying the embryonic mouse expression literature. The Gene Expression Literature Search (top left) allows querying of the embryonic expression literature for genes and ages analyzed and expression assay types used, as well as querying for bibliographic information or specific words in the title or abstract. A portion of the summary return for the query formulated in the figure is displayed at right. The table at the top, “Index Results by Age and Assay”, shows the number of matching records grouped by the age of the specimen and the assay type used. The lower portion of the page, “Index Results by Gene and Reference”, lists the citations for the references where the matching index results were reported, as well as the number of matching results contained therein. Entries marked with an * indicate they have been annotated in detail in GXD. Links on this summary page access detail pages (lower left). These pages display the expression information about the gene contained in the reference and provide links to gene and reference detail pages, as well as to the detailed expression data from the paper if they have already been annotated in GXD.