BACKGROUND: Anatomy ontologies play an increasingly important role in developing integrated bioinformatics applications. One of the primary relationships between anatomical tissues represented in such ontologies is part-of. As there are a number of ways to divide up the anatomical structure of an organism, each may be represented by more than one valid partonomic (part-of) hierarchy. This raises the issue of how to represent and integrate multiple such hierarchies. RESULTS: In this paper we describe a solution that is based on our work on an anatomy ontology for mouse embryo development, which is part of the Edinburgh Mouse Atlas Project (EMAP). The paper describes the basic conceptual aspects of our approach and discusses strengths and limitations of the proposed solution. A prototype was implemented in Prolog for evaluation purposes. CONCLUSION: With the proposed name set approach, rather than having to standardise hierarchies, it is sufficient to agree on a suitable set of basic tissue terms and their meaning in order to facilitate the integration of multiple partonomic hierarchies.

The Edinburgh Mouse Atlas Project (EMAP) has developed a digital atlas of mouse development which provides a bioinformatics framework to spatially reference biological data. The EMAP core database contains 3D grey-level reconstructions of the mouse embryo at various stages of development, a systematic nomenclature of the embryo anatomy, and defined 3D regions (domains) of the embryo. The reconstructions define a spatial framework for mapping data. Data front an in-situ gene-expression database is spatially mapped onto the atlas allowing the users to query gene-expression patterns using the 3D embryo model as a reference. The mouse atlas and gene-expression databases are publicly accessible through a set of Web-based tools. The system consists of a set of tools and databases, some of which reside locally on the Mouse Atlas hosts, others are remote. The middleware layer is primarily CORBA-based, but also makes use of Java servlets. Data are primarily stored in an object-oriented database system (ObjectStore). As a public bioinformatics resource, the Mouse Atlas system must be easily accessible to researchers all over the world, hence interoperability is a key issue. The Mouse Atlas is an on-going research and development project at the Medical Research Council, Human Genetics Unit, in Edinburgh. Access to its databases and further information is available through its Web site.

This paper provides a simple introduction to the reconstructions and data-handling tools stored on the Edinburgh Mouse Atlas CD, together with some of the ways in which the viewers and software can be used to understand mouse development and analyse data. The key aspect of the Mouse Atlas is that the underlying models are a complete representation of the histology, which has not been constrained to a particular interpretation. This means, for example, that the current anatomy domains can be further subdivided as required to any resolution up to the resolution of the models (2-7 microm). In the CD of the early embryos described here, virtually all tissues that can be usefully distinguished either by the histology or morphologically have been delineated.

An ‘undo’ capability is an interactive recovery facility provided by interfaces to enable users to reverse the effects of previously issued commands. The paper reviews the motivation offered in the literature for having such a facility; surveys existing practice in providing such a facility and formulates a set of guidelines based on a user-oriented research project to assist future practice in designing an undo capability for interfaces.

This paper generalises approaches to modelling an undo facility for interactive systems into a comprehensive user recovery and command reuse facility. It separates different undoing actions into distinct undoing functions and incorporates redoing capability in a more general command reuse capacity. Four adequacy criteria for such a facility are proposed and a general model is developed to meet these requirements. Partial, patterned and repetitive undoing and redoing actions are allowed on simple, complex and meta commands. The model subsumes the functionality of prior models.