I’m applying my Entropy Agglomeration algorithm to a dataset I found.
Here is a dendrogram that shows all experimented plants that interact via at least 5 genera of Mycorrhizal fungi (according to “MycoDB” database where all experiments on Mycorrhizal fungi in the literature are enumerated):
Bifurcations with lower entropies indicate that those plants’ interactions share common subsets of fungi. This one is based on all plant-fungus matchings with unit weights. It can be redrawn according to any additional weightings or restrictions put over the entire set of plant-fungus matchings. It can also be drawn to show the relations among the fungi, instead of the relations among the plants, using the same interaction data.
Mycorrhizal networks are also called Wood Wide Web.
Let me continue my little study on Mycorrhizal fungi to exhibit and demonstrate what one can dig in MycoDB.
This time I view the same data from the other perspective. The second dendrogram shows the relations among all genera of Mycorrhizal fungi that interact via at least three plant species:
Bifurcations with lower entropies indicate that those interactions incorporate larger common subsets of plant species.
Both plots are based on all plant-fungus matchings with unit weights, & any additional weighting or restriction can be easily incorporated into the procedure.
In each plot, the differences of one side is expressed in terms of the repetitions of the other side: When I’m interested in the differences among the plants, I reduce fungi to repetition. When I’m interested in the differences among the fungi, I reduce plants to repetition. The differences-of-interest are expressed in terms of “projection entropies” using REBUS, coded in Python.