3D developmental atlases are used in research for capture, collation and analysis of spatio-temporal data such as in situ gene-expression. The most advanced systems are based on a temporal series of 3D models. Examples are the EADHB human embryo atlas in Newcastle and the e-MouseAtlas in Edinburgh. Here we propose to use the 3D spatio-temporal frameworks in conjunction with a repository to deliver research and educational material directly in the context of the developing embryo. The primary interface will be similar in function to the familiar Google Maps but with full 3D indexing, spatial annotation and the overlay visualisation of complex spatial patterns. The project will deliver an integrated repository for embryology and developmental biology and open-source tools and APIs to enable institutions to build their own research and educational materials in the 3D framework.
§1. Embryo development involves a complex interplay of gene-expression involving spatial patterning and interactions to control morphogenesis (development of shape) and differentiation (development of tissues). Research in and teaching of development biology depends critically on understanding this spatial patterning within the embryonic structure and the dynamic changes of these structures. Conveying this understanding and making it accessible to students, researchers and the public is a very hard challenge. Today, the EADHB and the e-MouseAtlas provide the most advanced repositories for research data in the context of developmental biology. They contain spatial models for stages of development in the human and mouse respectively and a large database of gene-expression patterns and anatomy annotation. In addition, in conjunction with the Roslin Institute , models for the development of the chick are being added. It is not the mere structural referencing of anatomy that is important, but its potential to allow annotation of these spatio-temporal structures without relying on any particular anatomical knowledge. Prime examples are in situ gene expression results, which are continuously added to the repository via a rigorous curation process . This process is time consuming and a bottleneck to the expansion of content in the repository. Moreover, it is restricted to one particular type of experimental result, which does not utilise the atlases to their full potential.
§2. The e-MouseAtlas group in the Medical Research Council Human Genetics Unit in Edinburgh has extended the Image Internet Protocol (IIP) open standard to allow access and visualisation from 3D voxel images. A prototype web-browser based viewer has been implemented and delivers very fast “pan and zoom” viewing of arbitrary digital sections through the data (see Figure 1). This functionality has been delivered in the past by very high-performance parallel systems with architecture specific user interfaces. The merit of the MRC system is that very large image volumes (>20GB) can be delivered from standard server architectures and used in the context of web-browsers. This system is in regular use by large-scale projects running from the HGU and the whole system is open-source and available for free download.
§3. In this project we aim to enhance the e-MouseAtlas and EADHB repositories and image servers to enable other repositories to use it as a spatio-temporal reference framework. This idea is closely related to Google Maps and Google Earth, which are used by many other projects as the primary way to organise and present project-specific geographical information. We note the actual technology produced by Google is not directly useful in our context as it is tied tightly with geospatial systems; it uses only one model (i.e. the Earth), has no time component and is a 2D framework. In the context of developmental biology, the goal is to allow other repositories to link their data with the e-MouseAtlas by overlaying data entries from their repository on 3-dimensional models of embryos of all stages of development. The repository and server will be extended to include the EADHB models of human development and provide a unified resource.
§4. The extended repository will have three major groups of stakeholders: contributors, teachers, and learners. Contributors are currently the research groups generating the published and directly contributed research data that is the core of the e-MouseAtlas gene-expression database . This is curated therefore contributors are one step removed from the DB entries. The extended repository holds multi-media information (including linking remote data) with a spatial mapping to the framework can now be directly managed. The teaching stakeholders use the e-Atlas models either as downloaded data-sets or via cdroms that are distributed free of charge. The issue for this group is that data extensions and software upgrades are not easily managed and for teaching of larger groups rather impractical. A web-based solution is needed. For learners the enhanced repository will deliver rich content directly linked to the spatial framework of the embryo itself. For example the student query simply on the basis of location, learn the associated anatomy, link to tutorial material, advanced visualisations, processes and external data sources and media. The environment can also be used to test and monitor knowledge for example “draw the developing heart in-flow tract”.
§5. Figure 2 shows a typical developmental-biology teaching scenario. Here we show a student who queries the educational repository for material on the frontal lobe through a web browser. The repository shows a list of possible results found. After the student selects “future forebrain, mus musculus, stage TS14” from this list, the educational repository requests the corresponding model from the e-MouseAtlas, overlays reference other material available on spatial regions in this model and then pushes out the web-viewer with the model and the overlay. The student then can navigate this 3-dimensional model and select material from the repository. Here the student chooses a set of PowerPoint slides related to “frontal lobe development”. This story is similar to a scenario where a conference delegate locates the conference venue on Google Maps and queries for hotels, so as to pick a hotel nearby.
§6. Figure 3 shows the proposed architecture of the extended repositories. The original design with direct access just to the material of say the e-MouseAtlas is extended to enable additional material that is associated with the spatio-temporal framework to be assembled but not bound by the e-MouseAtlas resource itself. This critical step opens up the framework for widespread use as a standard to collect and collate teaching (and research) resources. We identify the four major elements required to improve the rate and diversity of material associated with the e-MouseAtlas.
§7. First, we need to enhance the e-MouseAtlas to allow external software to link into its 3-dimensional models of biological development. This will be achieved by building a web service that provides an interface to make use of the spatio-temporal framework. Essentially, this service will take in annotations and overlay these annotations on a view of a particular model. This model and overlay can then be embedded into a web page using the existing browser-based section viewer.
§8. Second, we need to enhance an existing repository software package to include the existing e-MouseAtlas web-viewer (shown in Figure 1) and the ability to overlay references to its material on these models. We have chosen DSpace as the repository software as it is the preferred option for many repositories worldwide, it is developed under an Open Source license, it is easily extended via a plug-in mechanism, and it fulfils many of the requirements JISC has identified as essential. We will develop a plug-in that given the metadata of a subset of its content, will connect to the e-MouseAtlas services to overlay this metadata on to a model and then allow the resulting model and overlay to be embedded on a web page. Simply put, this plug-in is another novel way of navigating a site’s content.
§9. Third, we need to populate this repository with educational material and to create a spatio-temporal link for each data item to the e-MouseAtlas. We will target a number of areas of developmental biology/embryology teaching that can demonstrate the power of this approach and for which the Newcastle group have close teaching links. These are heart and brain development and to test the repository local teaching material will be linked with the embryo models to show how the complex 3D structures develop. By linking in this way students will be provided with an integrated or systems view of the morphogenetic and differentiation processes that include standard developmental biology, spatial patterning and genetic regulatory control all in the context of the embryo. When content is added to the repository, a requirement is to choose which models and spatial areas this content is related with.
§10. Fourth, we need to integrate and promote the use of the educational material in existing developmental biology courses. The developmental biologist employed in Newcastle will liaise with the module leaders of the Developmental Genetics module in the Newcastle MSc Medical Genetics course to help them develop teaching materials for brain and heart in the context of the new resource. During the second year specific elements of the developmental biology course will be brought into the system and used to enhance the teaching. This will enable testing and tuning of the system in the context of real teaching.
§11. Once the system is running and tested with exemplar material it will be made available to the academic community as an active resource for direct use, as a repository to collect and collate training and tutorial materials and as an open source system for other Institutions to adopt.