Research training at Tübingen University with Daniel Huson
(En français)
Topics
Optimizing the layout of phylogenetic graphs.
To represent the evolution of some species or genes, the classical model is
the phylogenetic tree.
But trees are unable to represent horizontal gene transfers,
recombinations and hybridizations. In those
cases,
phylogenetic networks are required.
The
SplitsTree software
can represent many of those networks, in particular
splitsgraphs.
Those graphs represent
conflicting signals in a phylogeny,
with
boxes appearing wherever
evolution is uncertain. A splitsgraph can be obtained by a bootstrap on
phylogenetic trees, or by a consensus
of different phylogenetic trees, even if they
have a different number of taxa.
My purpose in this research training is
to improve the layout of these splitsgraphs in SplitsTree,
by increasing the area of the boxes which represent
the conflicting signals in the phylogeny.
I worked on this problem with Daniel Huson.
Optimizing the layout of reticulate networks.
When
reticulations are added to a phylogenetic tree, the new edges created
may intersect with existing edges. But changing the ordering of the taxa
could lessen the number of such intersections. With Tobias Klöpper, our purpose was to
find a heuristic to
determine the ordering on the taxa which would minimize
the total number of crossings. The one we propose is based on the
dual graph of the phylogenetic tree,
with a distance which reflects the number of
intersections, and a
mirroring process
to improve this distance in a
simulated annealing scheme.
Results :

10 loops of the "box-opening algorithm".
Back to the main page