• The conceptual framework which houses the descriptions of the expression patterns in EMAGE is the EMAP Mouse Embryo Anatomy Atlas.
  
  This consists of a set of 3D virtual embryos at different stages of development, as well as an accompanying ontology of anatomical terms found at each stage:

• Sites of gene/enhancer expression in EMAGE are described by denoting appropriate regions in the EMAP virtual embryos where expression is detected (and not detected) and also describing this information with an accompanying text-based description, which is achieved by referring to appropriate terms in the anatomy ontology:

EMAGE Data Annotation Methods

The upper panel shows two examples of WM data annotation. Both show right-hand side views of WM stained embryos at TS15 (~9.5dpc). These have both spatially annotated to the corresponding right-hand side view of the TS15 model. The different colours represent different levels of expression (e.g. red = strong, yellow = moderate, and cyan = no detectable expression).

The lower panel shows two examples of section data annotation. Both depict views of sections (either 1 or 3 sections) taken from TS12(~8.0dpc) embryos. These have been spatially annotated onto an appropriately matching section plane chosen from the TS12 virtual embryo. Views of the spatial annotations are shown here directly on the virtual section planes and in the 3D context of the virtual embryo model.

The accompanying text annotations to terms in the EMAP anatomy ontology (right-hand column), are based on the original descriptions given by the author. Larger image.

• Spatial mapping is used to warp raw image data onto a stage-matched model. This allows the EMAGE Editorial Team to spatially map gene expression patterns.

EMAGE 3D Spatial Warp

The upper panel shows points of equivalence or landmark points that are manually added to 3D rawimage data and a stage-matched 3D model. These landmark points are used to spatially warp the raw image data onto the 3D model. This warping procedure utilises a Constrained Distance Transform (CDT) and allows the EMAGE Editorial Team to spatially map gene expression patterns.

The middle panel highlights the steps that are used to generate a space-filling tetrahedral mesh that is utilised by the CDT. A domain is derived from the embryo model, but not the surrounding space, and this is used to generate a 3D mesh that accurately describes the 3D model. Internal spaces in the embryo, such as the neural lumen of the developing brain, are also omitted from the mesh. The mesh serves to constrain the spatial warp, allowing for an organic deformation of raw image data.

The lower panel shows raw OPT data and this gene expression pattern spatially mapped onto a grayscale model.