Monday, November 11, 2013
Presenting complex splits graphs
One common problem when presenting results using a data-display network is the complexity of the relationships among the samples, especially when there is a large number of them. It is often the case that the relationships among closely related samples are impossible to see clearly.
A recent paper (El Baidouri F, Diancourt L, Berry V, Chevenet F, Pratlong F, Marty P, Ravel C (2013) Genetic structure and evolution of the Leishmania genus in Africa and Eurasia: what does MLSA tell us. PLoS Neglected Tropical Diseases 7: e2255) presents an interesting solution to this problem. Basically, the idea is to present a series of graphs, with the main graph showing the overall relationships and a collection of small graphs showing the details of different parts of the network.
This takes longer, of course, as it involves doing a series of analyses, one for each subset of the data, but this is easy enough to do in programs like SplitsTree. It seems to be an idea worth considering.