Introduction to Bioinformatics

1
Introduction to Bioinformatics

• Phylogenetics
• Part III
• Character-Based Methods

2
Distance Methods

• Complexity
• Distance-based methods much faster than other
  methods
• Commonly used in multiple sequence alignment
• UPGMA PILEUP
• Neighbor-joining CLUSTALW http//www.ebi.ac.uk/c
  lustalw/
• Problems
• Both UPGMA neighbor-joining are greedy
  heuristics
• Possible to be trapped in local maxima (no
  backtracking)
• Output is a single tree, even if many equal-cost
  alternatives
• Big Picture only
• May use as starting point
• Tree generated provides upper bound for
  branch-and-bound
• Initial tree for probabilistic branch-swapping
  techniques
• Other approaches?

3
Character-Based Methods

• Maximum Parsimony (MP) Fitch 1971
• Minimize number of sequence changes in tree
• Assume fewest changes (mutations) most likely
  (evolution)
• Informative site
• Position with useful change information (for
  parsimony)
• I.e., of changes in position dependent on tree
  chosen
• Must have 2 different bases / residues, such
  that each base / residue appears in 2 sequences

4
MP Example

• What are the informative sites in this example?
• Build distance matrix

5
(No Transcript)
6
MP Example

• Most parsimonious tree
• Tree with fewest total of changes at
  informative sites
• Continue with our example
• Informative sites
• Seq1 GG
• Seq2 GT
• Seq3 AG
• Seq4 AT
• Site changed
• Tree 1 __
• Tree 2 __
• Tree 3 __
• Which tree?

7
MP Method

• Algorithm
• Generate all possible tree topologies
• Count number of changes required
• Select tree with minimum changes
• Use branch-and-bound to reduce search
• Search trees with increasing of leaves
• Abandon subtree when changes best completed
  tree
• Characteristics
• Computationally expensive
• Analyze only informative sites
• Misleading if rates of changes vary among
  branches
• Evolution is not always parsimonious

8
MP Method

• Can infer ancestors
• An internal node
• Intersection of two children, if it is not empty
• Union of two children, otherwise
• unions substitutions

(GT)
9
Tree Construction Issues

• Selecting tree construction algorithm
• If strong sequence similarity ? maximum parsimony
• If clearly recognizable sequence similarity ?
  distance methods
• Otherwise ? maximum likelihood
• Determining statistical significance
• Multiple tree shapes possible
• Find probability that tree shape is as described
• Sample by bootstrapping Efron Tibshirani
  1993
• Generate artificial data set by repeatedly
  selecting random columns of alignment
  (pseudo-alignment) with replacement
• Build tree for pseudo-alignments many (1000)
  times
• Frequency phylogenetic feature appears ?
  confidence level

10
Tree with Bootstrap Values
Source http//fungal.genome.duke.edu/images/fungi
_subset_tree.jpg
11
Phylogenetics Issues

• Gene trees vs species trees
• Gene duplication can complicate phylogenetic
  analysis
• Paralogues (duplicated genes) do not fit in
  evolutionary tree
• Choice of target sequence type
• Ribosomal RNA (slowest change / mutation rate)
• Use for very long-term evolutionary studies,
  spanning species boundaries biological kingdoms
• DNA / RNA (fastest change / mutation rate)
• Use for short-term studies of closely-related
  species
• Contains more evolutionary information than
  protein
• Protein (medium change / mutation rate)
• Use for wide species comparisons
• More reliable alignment than DNA

12
Phylogenetics Summary

• Phylogenetic prediction
• Infer evolutionary relationships from shared
  features
• May have application to sequence alignment,
  epidemiology
• Phenotypic vs. genetic (i.e., molecular)
  characteristics
• Phylogenetic trees
• May be ultrametric and / or additive
• Tree construction
• Inexpensive distance-based (UPGMA,
  neighbor-joining)
• Expensive (exhaustive) tree searches (parsimony,
  likelihood)
• Assessing phylogenetic trees
• Algorithms always produce some tree (of varying
  accuracy)
• Expert biology knowledge to assess correctness /
  significance

13
Phenetics vs Cladistics

• Phenetics uses all the data
• Uses overall similarity to group taxa not
  necessarily evolutionary 
• Any kind of object could be subjected to a
  phenetic analysis 
• Taxa that are more similar are grouped together 
• Cladistics uses only informative sites
• Taxa are grouped together based on patterns of
  sharing of derived character states
• Taxa sharing a derived character state do so
  because they inherited this character state
  through a common ancestor
• Advantages of the cladistic approach
• Less susceptible to such rate variation
• Shared, derived character states won't mislead
  you
• Birds are dinosaurs cladistic perspective

14
Applications Building Tree of Life
15
Applications Building Tree of Life
Source http//www.isem.univ-montp2.fr/PPP/PM/RES/
Phylo/Mamm/PHYLMOL-Placentalia7EEnglish.jpg
16
Application - CSI

• Which patients are more likely to be infected by
  the dentist?

Source http//trc.ucdavis.edu/djbegun/Lect_12.1.h
tml
17
Application -Human migration

• Based on mtDNA genome
• Africans have twice as much diversity among them
  as do non-Africans ? Africans have a longer
  genetic history
• More recent population expansion for non-Africans
• Africans and non-Africans diverged recently
• Out of Africa

Source Discovering Genomics, Proteomics,
Bioinformatics, by Campbell Heyer
18
Software

• http//evolution.genetics.washington.edu/phylip/so
  ftware.html
• 301 of the phylogeny packages
• 39 free servers