Notes 20100713.122620 CS 798 Phylogenetic Networks Evolutionary Studies

From SnOwy - Ed's Wiki Notebook

Contents 1 2006 2 Why phylogenetic networks? 3 More reasons for phylogenetic networks 4 Split networks

2006

• Survey paper

Why phylogenetic networks?

• Why not trees in all cases?
  1. Evolution is non-tree-like
  2. Tree-like, but there are inconsistent signals due to...
     • model error
     • sample error
• Why networks?
  1. is a sufficient model
  2. is a more general case
  3. split networks: a summary argument
  4. fitness of the model as a network
     • tests consistency

More reasons for phylogenetic networks

• reticulation (web-like)
  • internal nodes reflect hypothetical ancestors
• coalescence (different strain recombination)
• able to deal with phylogenetic trees

Split networks

• useful for summarization and fitness
  • internal nodes do not reflect hypothetical ancestors necessarily
  • inconsistent data creates split network internal nodes which we infer to be a signal problem

A split network

       |  |
       *  |
       |\ |
       | \|
       |  *
       *  |  /
    \  |\ | /
     \ | \|/
      \|  *
       *  |
        \ |
         \|
          *

• kinds of Split networks...
  • Median
    • rom sequence data
  • Split decomposition
  • Neighbour networks (from NJ → network instead of tree)
    • from distances
  • Concensus net supernetworks
    • from trees
• split - bipartition (non-empty bipartition)
• notation: {A | B} ∈ τ
• Split system - set of splits
• Split graph - graph displaying each split in the split system
• A split graph implies a unique system of splits
  • however, a unique system of splits does not guarantee a unique split graph
• keywords: compatible split, conflicting split
• compatible split system: every pair of splits is compatible with all other splits -- produces a tree
  • if system is incompatible: produces a network
• Given a set of trees generated elsewhere which all claim to have the same correctness value ...
  • get the set of bipartition splits (implying the trivial trees)
  • count for each split the number of occurrences in the set of input trees
  • The usefulness of this summary method allows us to express concensus
  • these splits represent conflicts in the data