Figure 8. Prototype mock-up.

Bar charts of attributes are shown in (A) - in this case classes of GO terms. These bar charts detail all attributes associated with the gene products displayed in the network view, shown in (B). In (A), bar length indicates the number of times an attribute occurs as a trait of the gene products in the network. When a scientist selects 2+ attributes from the bar chart, the associated gene products in the network (B) and matrix (C) are dynamically highlighted. Additional attributes in (A) that co-annotate the highlighted proteins also are dynamically highlighted. The matrix cells in (C) show the attributes (rows) that annotate each gene (columns). By interacting with the matrix, a user can filter to only certain attributes, and the views will show only the genes having those attributes of interest. Results dynamically update across the graph, bar charts, and data table (D). A user can set various visual codings to indicate traits of most interest and can manually group gene products in the network that share two or more traits. The user can view the strength of interactions in potentially interesting groupings by referring to the details on edges in the data table. These four dynamically linked views should foster a fluent and complete flow for the objective of selecting and grouping by 2+ similarities.