Today, I would like to mention about the graph abstraction and visualization with several levels. We are now able to visualize the huge graphs at one window. There are some examples which were posted before like the Internet Graph. However, when you look at the whole layout, you could not easily detect useful semantics which is too complicated to understand. So we need an abstraction or visualize within the levels methodology in order to give both details and general perspective.
Abstraction can be done in several ways. One way is clustering. Clustering may occur due to the nature of the graph or it may be done intentionally. One of the widely used algorithm is MCL(Markov Clustering Algorithm) and publicly available. This clustering can be useful to divide the graph into pieces. Furthermore, one can use some filtering methodologies to relax the layout production and view.
Also, visualization within levels is also important. When you level the problem, you can use different layouts for different levels in order to make the visualization semantically. In Robinviz we have done this methodology. We proposed a visualization solution for biclustering using another paradigm called Protein-Protein Interactions network. One central view shows the relations between biclusters and peripheral views shows the corresponding PPI graph.
It is also interesting fact that biological interaction data has also several levels. PPI's are one aspect. The lower aspect molecular interactions which I am now working on their visualizations, is also at one level depth.
Anyway, the total visualization system for these interactions using graph drawing is exciting research that could be perfect for biologist :). However, it is really though business.