SplitsTree

1
SplitsTree
Vincent Moulton The Linnaeus Centre for
Bioinformatics
2
Tree of Life.
(M.Sogin D.Patterson, Tree of Life Web
project)
3
..or network ?
4
Why use networks to analyse evolution?

• Tree can be an inappropriate evolutionary model

• Visualization of complex evolutionary patterns

• Data exploration

5
Distance methods
6
Quartets
Can represent a distance matrix on four taxa
A,B,C,D by weighting the following graph
7
Example
8
Tree-likeness
9
plots (Holland, Huber, Dress, Moulton MBE 2002)
d
10
Example

• 42 isolates of yeast C.albicaus
• Distances computed from AFLP bands
• Clonal (tree-like) vs sexual (nontree-like)
• reproduction?

11
Quartet-mappings (Strimmer, von Haeseler, PNAS,
1997, Nieselt-Struwe, von Haeseler, MBE 2001)
12
Example
13
Detecting Recombination
14
Highway Plots (Strimmer, Forslund, Holland,
Moulton, Gen. Bio, 2003)
15
HIV genome scan
Data from Piyasirisilp et al., Journal Virology,
2000
Generated with VisRD software package, Forslund,
Huson, Moulton, submitted
16
How do we get networks from distance matrices?
?
17
Splits
c
d
b
a
e
A a,b B c, d, e   Denote by A B
18
Compatible splits
c
d
b
e
a
a, b c, d, e and a, b, c d, e are
compatible
19
Fact (Buneman 1971 and others)
A collection of splits of a set of taxa in which
each pair of splits is compatible corresponds to
a unique tree labeled by the taxa.
20
Problem
Given n taxa there are 2(n -1) 1 possible
splits, so how should one find a relevant
collection of (compatible) splits?
21
Splits from distances
c
b
3
2
3
1
1
3
1
d
4
a
1/2 (ac bd (ab cd)) 1/2 (11 7
(7 5)) 3
22
Split decomposition(Bandelt and Dress, 1992)
A
D
F
C
E
B
G
Isolation index of the split ABCD EFG equals
minimum box side length taken over all boxes
crossing the split
23
Isolation index

• Define
• (ab cd) max ac bd, ad bc (ab cd)

• Given a split A B define
• (A B) 1/2 min ?(ab cd) a, b in A and c,
  d in B

24
Fact (Bandelt/Dress 1991)
Given any distance on a set of taxa, the
collection of splits A B with ? (A B ) gt 0 is
weakly compatible.
25
Consequence
The collection of splits that positive isolation
index is not too large and can be efficiently
computed.
26
Representing weighted splits by graphs
AB CDEFG, 3 AF BCDEG, 2 CF ABDEG,
1
A,B,C,D,E,F,G
27
Split graphs
0.1
0.05
0.2
0.1
0.2
0.1
0.5
0.3
0.1
0.1
0.2
0.2
0.1
d(B,D)
1.35
28
Summary
29
Middle Earth
30
Color circle
31
Virus data DEN 1
32
Systematic bias
A
D
F
C
E
B
G
33
Result loss of resolution
(mitochondrial gene order data from early
branching Eukaryotes using normalized breakpoint
distance)
34
Neighbor-Joining (NJ) (Saitou/Nei 1987)

• Happy to give a tree, whatever the data.

• The most widely used distance based phylogenetic
  method.

35
What do you get if you cross SplitsTree with
Neighbor-Joining?
?
36
Neighbor-Netting(Bryant and Moulton, MBE 2003)
37
Example
38
SplitsTree 3.2
Program for windows written by Daniel
Huson. http//www-ab.informatik.unituebingen.
de/software/splits/welcome_en.html
39
SplitsTree 4
All-new implementation of SplitsTree in Java,
developed by Daniel Huson and David Bryant.
http//www-ab.informatik.uni-tuebingen.de/software
/splits/welcome_en.html
40
Summary

• SplitsTree provides an easy to use tool for
  exploratory data analysis.

• It provides a means to help visualize the
  complexity in
• phylogenetic data.

• The extent and localization of incompatibilities
  within split graphs
• may be alerting you to something
  interesting about underlying
• biological processes.

• SplitsTree can help inform you of the
  suitability of your data for
• building optimal bifurcating trees.