Table 1 Outline of solutions employed by the iRefScape plug-in

Data content and delivery

Multiple source databases with redundant and non-homogenous data.

Consolidation and normalization of data.

1.1

Need to retrieve a large dataset for subsequent visualization and filtering. Need for a private and static data set.

Operate on a local data set.

Use of file system and collision-proof identifiers for storage and retrieval of meta-data.

1.2

Search

Need to support searching on multiple accession systems for proteins.

Use of universal ROGIDs and RIGIDs.

2.1

Differential choice of splice isoforms when curating interaction data.

Use of canonical expansion during search.

2.2

First order neighbours do not accurately reflect the neighbourhood of a protein.

Use of neighbourhood completion for query results.

2.3

Visualization

Need to track search result provenance.

Use of features and colour to indicate result level. Use of "Grid" layout for initial results.

3.1

Need to visualize all evidence supporting a given interaction.

Use of a multigraph representation

3.2

Need to distinguish between binary and n-ary interaction data.

Use of bipartite model to visualize n-ary data.

3.3

Need to view/filter search results on multiple node and edge attributes.

Distribution of rich and normalized attribute set.

Advanced filter function.

Dedicated "between-nodes" search.

3.4

Data discovery and navigation

Need to identify potential spoke-represented complexes.

Dedicated search tool.

4.1

Need to search on groups of genes related to a disease.

Dedicated disease-group search.

4.2

Need to explore connection between two groups of genes.

Use of synchronized adjacency-matrix viewer.

4.3

Need for interoperability with other plug-ins.

Export functionality.

4.4

Need for user to navigate a complex interface.

Use of wiki-documentation, in-line help and macros.

4.5