The Genome Access Course Phylogenetic Analysis

1
TheGenomeAccessCoursePhylogenetic Analysis
2
The Great Chainof Being
From Didacus Valades, Rhetorica Christiana (1579)
3
Phylogenetics

• Developed by Willi Henning (Grundzüge einer
  Theorie der Phylogenetischen Systematik, 1950
  Phylogenetic Systematics, 1966)

4
What is the ancestral sequence?

• pfeffer
• pepper
• (pf/p)e(ff/pp)er

5
What is the ancestral sequence?

• five
• quinque
• pento
• funf
• panj
• viisi
• pompe
• queig
• pindzé

penkwe
6
What is the ancestral sequence?

• HGLCSAIP
• HGICSGIP
• HG(L/I)CS(A/G)IP

7
Evolutionary Trees

• A tree is a connected, acyclic 2D graph
• Leaf Taxon
• Node Vertex
• Branch Edge
• Tree length sum of all branch lengths
• Phylogenetic trees are binary trees

8
A Generic Tree
9
Evolutionary Trees

• Rooted
• common ancestor
• unique path to any leaf
• directed
• Unrooted
• root could be placed anywhere
• fewer possible than rooted

10
Rooted Tree
generated by DRAWGRAM (PHYLIP)
11
Unrooted Tree
generated by DRAWTREE (PHYLIP)
12
Possible Evolutionary Trees
13
Paralogs Orthologs
1A
2A
3A
1B
2B
3B
14
Genes vs. Species

• Sequences show gene relationships, but
  phylogenetic histories may be different for gene
  and species
• Genes evolve at different speeds
• Horizontal gene transfer

15
(No Transcript)
16
(No Transcript)
17
Methods for Phylogenetic Analysis

• Character-State
• Maximum Parsimony
• Maximum Likelihood
• Genetic Distance
• Fitch Margoliash
• Neighbor-Joining
• Unweighted Pair Group

18
Phylogenetic Software

• PHYLIP
• PAUP (Available in GCG)
• TREE-PUZZLE
• PhyloBLAST
• Felsenstein maintains an extensive list of
  programs on the PHYLIP site

19
PHYLIP Programs

• dnapars/protpars
• dnadist/protdist
• dnaml (use fastDNAml instead)
• neighbor
• fitch/kitsch
• drawtree/drawgram

20
Maximum Parsimony

• Most common method
• Allows use of all evolutionary information
• Build and score all possible trees
• Each node is a transformation in a character
  state
• Minimize treelength
• Best tree requires the fewest changes to derive
  all sequences

21
Which is the more parsimonious tree?
9 Node Crossings
8 Node Crossings
22
Maximum Likelihood

• Reconstruction using an explicit evolutionary
  model
• Tree is calculated separately for each nucleotide
  site. The product of the likelihoods for each
  site provides the overall likelihood of the
  observed data.
• Demanding computationally
• Slowest method
• Use to test (or improve) an existing tree

23
Clustering Algorithms

• Use distances to calculate phylogenetic trees
• Trees are based on the relative numbers of
  similarities and differences between sequences
• A distance matrix is constructed by computing
  pairwise distances for all sequences
• Clustering links successively more distant taxa

24
DNA Distances

• Distances between pairs of DNA sequences are
  relatively simple to compute as the sum of all
  base pair differences between the two sequences
• Can only work for pairs of sequences that are
  similar enough to be aligned
• All base changes are considered equal
• Insertion/deletions are generally given a larger
  weight than replacements (gap penalties).
• Possible to correct for multiple substitutions at
  a single site, which is common in distant
  relationships and for rapidly evolving sites.

25
Amino Acid Distances

• More difficult to compute
• Substitutions have differing effects on structure
• Some substitutions require more than one DNA
  mutation
• Use replacement frequencies (PAM, BLOSUM)

26
Fitch Margoliash

• 3 sequences are combined at a time to define
  branches and calculate their length
• Additive branch lengths
• Accurate for short branches

27
Neighbor Joining

• Most common method of tree construction
• Distance matrix adjusted for each taxon depending
  on its rate of evolution
• Good for simulation studies
• Most efficient computationally

28
UPGMA Unweighted Pair Group Methods Using
Arithmetic Averages

• Simplest method
• Calculates branch lengths between most closely
  related sequences
• Averages distance to next sequence or cluster
• Predicts a position for the root

29
Phylogenetic Complications

• Errors
• Loss of function
• Convergent evolution
• Lateral gene transfer

30
Validation

• Use several different algorithms and data sets
• NJ methods generate one tree, possibly supporting
  a tree built by parsimony or maximum likelihood
• Bootstrapping
• Perturb data and note effect on tree
• Repeat many times
• Unchanged 90, trees correctness is supported

31
Are there bugs in our genome?
N-acetylneuraminate lyase
32
The End