All trees are rooted– think of a tree as being a collection of directed edges. Build graphs. Nodes are descendant sets, and edges are tree edges.
Combine all of the bootstrap trees into a graph. Can also combine morphology and taxonomy.
“Synthesis”– order types of trees according to reliability.
Likelihood estimate based on evolutionary simulation of population on its fitness landscape.
Here evolutionary models = trait values.
Parallelize nhML implementing GG98 model using MPI and OpenMP. They get nearly linear speedup with number of cores.
Pipeline– map using Bowtie2 or Blat with “sloppy parameters. Extract mapped reads into target bins Assemble each target individually Replace targets with assembled contigs Repeat.