Constructing Synthetic Phylogenies

1
Constructing Synthetic Phylogenies

• by Alex Kovarsky
• Project Presentation of C606

2
Outline

• Motivation
• Process Description
• Reconstruction and Comparison Algorithms
• Experimental Results
• Conclusion

3
Motivation
4
Phylogeny - What it is?

• The study of the evolution of life forms
• Tree of life forms and their relationships
• The history of lineages as they change through
  time

5
Real Phylogenies

• Estimate 5 to 100 million of organisms living
  today on earth
• All organisms are connected by the passage of
  genes along the branches of the phylogenetic tree
  
• In phylogenies we can find similarities and
  differences in plants, animals, etc.
• Tree of Life Web Project (ToL)1
• Collaborative world wide project
• 3000 web pages containing information about
  particular groups of organisms
• ToL pages are linked to one another
  hierarchically

6
Evolutionary Process

• Main instrument of evolution is mutation
• Organisms with novel traits are created
• Organisms with mutated alleles migrate and mate
  in order to create variation in the population
• Forces of natural selection then choose the
  better-adapted organisms
• However, most mutations are harmful

7
Synthetic Phylogenies

• Phylogenies created by pseudo-random mutation of
  a single organism located in the root of the
  phylogenetic tree
• Conditions
• Mutation Rates
• Mutations Usability
• Gene Sizes
• Are based on current knowledge of evolutionary
  processes

8
Project Goals

• By applying known evolutionary knowledge to
  generate synthetic phylogenies
• Can learn about the process of evolution
• Many phylogeny inferring algorithms exist
• Need a way to test their accuracy and efficiency
• Real phylogenies are not suitable since we do not
  know the tree topology
• Thus, can use synthetic phylogenies

9
Method Description
10
Method General Description

• Start with a randomly generated genome made up of
  N genes
• This genome mutates creating new organisms
• Some of those organisms survive and multiply,
  while most do not
• This process is run for N cycles to create a
  phylogenetic tree of species

11
Mutation Types

• Point common
• Replacement
• Insertion
• Deletion
• Duplication - rare
• Translocation very rare

12
Point Mutations

• Replacement
• Changing of a random nucleotide in a random gene
  from 0 to 1 or vice versa
• Insertion
• Insertion of a random nucleotide in a random
  position of a random gene
• Deletion
• Deletion of a random nucleotide in a random
  position of a random gene

13
Rare Mutations

• Duplication
• Production of an exact copy of a random gene in a
  given genome
• Translocation
• Occurs when two neighbouring genes swap places in
  the genomic sequence

14
Harmful vs Beneficial

• Vast majority of mutations are
• Harmful
• have detrimental affects on the fitness of the
  organism to the environment
• Mutated genes will not transfer to the general
  population
• Only few mutations improve the organisms chances
  to survive
• Beneficial Mutations

15
Stable vs Evolving Population

• Stable
• Species whose population size is large enough to
  have a good likelihood to produce beneficial
  mutations
• Evolving
• Species whose population size is not high enough
  to have a good likelihood to produce beneficial
  mutations
• The typical trend
• Evolving ? Stable ? Extinct

16
Population Growth Process
01101110..11
Size 1, Growth 0.1
17
Population Growth Process
01101110..11
Size 1, Growth 0.1
01101110..11
01101111..11
Size 0.95, G -0.05
Size 0.08, G -0.2
18
Population Growth Process
01101110..11
Size 1, Growth 0.1
01101110..11
01101111..11
Size 0.95, G -0.05
Size 0.072, G -0.2
01101110..11
01001110..11
Size 0.86, G -0.1
Size 0.12, G 0.2
19
Population Growth Process
01101110..11
Size 1, Growth 0.1
01101110..11
01101111..11
Size 0.95, G -0.05
Size 0.072, G -0.2

01101110..11
01001110..11
Size 0.86, G -0.1
Size 0.12, G 0.2


20
Population Growth Process
01101110..11
Size 1, Growth 0.1
01101110..11
01101111..11
Size 0.95, G -0.05
Size 0.072, G -0.2

01101110..11
01001110..11
Size 0.86, G -0.1
Size 0.12, G 0.2


01101111..11
Size 0.009, G -0.2
01101110..11
01001110..11
Size 0.009, G -0.1
Size 0.3, G 0.1
21
Population Growth Process
01101110..11
Size 1, Growth 0.1
01101110..11
01101111..11
Size 0.95, G -0.05
Size 0.072, G -0.2

01101110..11
01001110..11
Size 0.86, G -0.1
Size 0.12, G 0.2


01101111..11
Size 0.009, G -0.2
01101110..11
01001110..11
Size 0.009, G -0.1
Size 0.3, G 0.1
01001110..01
01001110..11
Size 0.08, G -0.2
Size 0.285, G -0.05
22
Reconstruction Programs

• All come from PHYLIP website
• PHYLIP is a free package of programs for
  reconstructing phylogenies
• Programs used
• PARS
• MIX
• PENNY
• DOLLPENY

23
PARS

• Executes Wagner Parsimony
• Finds the tree which requires the minimum number
  of changes
• Assumptions
• Ancestral states are unknown.
• Different characters evolve independently.
• Different lineages evolve independently.
• Changes to all other states are equally probable

24
MIX

• Executes the Wagner and Camin-Sokal parsimony
  methods in mixture
• Assumptions similar to PARS
• Except assumes that changes from 0 ? 1 are more
  probable than changes from 1 ? 0
• Not the most suitable method for reconstruction
  of our tree

25
PENNY and DOLLPENNY

• PENNY
• Uses branch and bound method
• Very slow
• DOLPENNY
• Uses branch and bound method
• Asserts that in evolution it is harder to gain a
  feature than to lose it
• 1 ? 0 more common than 0 ? 1
• Extremely slow

26
Tree Comparison

• Based on comparing the distances between all
  pairs of leafs in one tree to all pairs of leafs
  in the other tree
• Process
• Find all distances between leafs in both trees
• Normalize to (0,1 for both trees
• Compute the sum of distances in the generated
  tree
• Compute the sum of differences between all
  respective leaf pairs
• Similarity ratio is
• Sum of Differences4 / Sum of Distances 3

27
EXPERIMENTAL RESULTS
28
Experiments - Goals

• Examine performance of inference methods
• PARS, MIX, PENNY, and DOLPENNY
• Test the properties of created phylogenies
• Variability
• Growth rates

29
Experiment 1

• Performance of Reconstruction Methods on Simplest
  Mutations
• Only insertion and duplication mutations
• PARS method is the best performer

30
Experiment 2

• Performance of Reconstruction Methods on Complex
  Mutations
• Point Mutations, Duplication, and Translocation
• Much poorer results
• PARS method is the fastest performer

31
Experiment 3

• Variance of the current species size
• Randomization has a substantial affect on
  progress of evolution

32
Experiment 4

• Varying the Number of Current Cycles
• Suggests exponential growth in the number of
  species

33
Conclusion

• A tool that produces discrete synthetic
  phylogenies, according to user specified
  parameters
• All parameters are adjustable to accommodate
  quick modifications
• A test-bed for phylogeny inference methods

34
QUESTIONS?
35
References

• Maddison, D. R. and K.-S. Schulz (ed.) 2004. The
  Tree of Life Web Project. Internet address
  http//tolweb.org
• PHYLIP - http//evolution.genetics.washington.edu/
  phylip.html