Notes 20100708 CS 798 Quartet Phylogeny Reconstruction, Set Programming, Fathiyeh
From SnOwy - Ed's Wiki Notebook
Contents |
Quartet-Based Phylogeny
- Leaves and subtrees are considered as quartets linked such that there is either one or two nodes between any two nodes.
- We ask: what if the quartet topologies conflict with one another?
Maximum Quartet Consistency
- Exact algorithms
- Near-optimal solutions
- general approach
- translate the problem of searching for best quartet phylogeny into an ultrametric phylogeny problem
- each node is labelled with a positive value such that deeper nodes (closer to leaves) are smaller
- how do we perform this transformation?
- two theorems together show that we can perform this transformation under some circumstances AND that the runtime is O(n2)
- new form of logic programming; solving the problem in a declarative way: what, not how.
Inputs, Constraints, Goals
- Constraints:
- Symmetry, Ultrametricity, Quartet Consistency
Biologist's Questions
- Does this method of mapping the best quartet phylogeny onto an ultrametric matrix problem actually fare against existing methods?
- In the big picture, do we actually put each quartet back into a big phylogenetic where quartet nodes may be complete subtrees?
- Yes
Optimizations
- Hypercleaning: considers each edge and recovers the edges that have the best probability of being correct?
- Exponential time :(
- behaves better than existing algorithms: dynamic programming, etc.
What's the point?
- maximum quartet consistency is a reasonable way to look at phylogenetic reconstruction:
- because we aren't likely to find a phylogeny that will describe all of the relationships between taxa, we aim to find one that describes the most in a reasonable amount of time.