BIN6002 project report

1
BIN6002 project report
Phylogenomics based on a collection of
mitochondrial proteins

• By Tetsu Ishii and Tan Wang
• June 25, 2004

2
(No Transcript)
3
Our project task

• We are given a dataset of mt genome sequences
  from over 100 species.
• Through phylogenetic analysis (phylogenome
  approch), we want to place two species, nuclearia
  and emiliania in the previous phylogenetic tree.

4
Our dataset
103 mitochondia and a-proteobacteria genome from
different species
5
Which species and which genes were included in
our analysis ?

• Cob and cox1 gene were used to do a quick
  phylogenetic analysis using distance methods,
  species with long branch were eliminated.
• Some species were eliminated simply because we
  are not interested.
• Totally, 35 species including outgroup were
  selected
• The common genes set 11 genes were selected
  because they present in all remained species.
  These 11 genes were concatenated and used in the
  phylogenetic analysis.

6
The five steps in phylogenetic analysis
1
Sequence data
2
Align Sequences
Phylogenetic signal?evolutionary processes?
3
Distances methods
Characters based methods
Distance calculation (which model?)
4
Choose a method
phyml
protpars
Puzzle
Bionj
ML
MP
MB
Wheighting? (sites, changes)?
Model?
Calculate or estimate best fit tree
5
Test phylogenetic reliability
Modified from Hillis et al., (1993). Methods in
Enzymology 224, 456-487
7
Which methods should we choose ?

• We simply try all 5 methods, which we can access
• Maximum likelihood phyml, puzzle
• Distance method Bionj
• Maximum parsimony phylip (protpars)
• MrBayes

8
Which model should we choose ?

• Models are described in terms of the tendency of
  one amino acids/base to change to another
  site-to-site rate variation the composition.
• Dont assume a model. Rather, find a model that
  fits your data.
• We try both JTT with/without gamma distribution.

9
Two examples models
Site rate heterogeneity were modeled with gamma
distribution
10
Is the JTT model with gamma distribution better
than the JTT with no gamma for these data
(puzzle) ?

• model ln likelihood ?
• JTT w/gamma -76158.97
• JTT w/o gamma -82133.12 5974.15

2d2 (Inw/gamma-Inw/o gamma)
P0
11
puzzle
JTT w/gamma
ln likelihood -76158.97
12
puzzle
JTT w/o gamma
ln likelihood -82133.12
13
100
100
a-Proteobacteria
100
100
Rhodophytes
100
100
47
PlantsChlorophytes
48
100
69
100
Jakobids
88
100
100
Haptophyceae
Distance method
1
Cercozoa
100
90
59
100
Stramenopile
91
JTT w/gamma
98
Mycetozoa
92
61
93
Holozoa
97
100
100
100
100
Fungi
99
14
100
Cercozoa
100
Stramenopile
100
100
Maximum likelihood
Haptophyceae
40
PlantsChlorophytes
100
Jakobids
100
26
JTT w/gamma
100
Rhodophytes
100
100
27
Jakobids
100
Mycetozoa
98
86
100
87
Holozoa
86
100
100
100
Fungi
100
100
100
100
a-Proteobacteria
100
15
The conclusion

• With the phylogenome approach, we are able to
  place nuclearia within Holozoa and emiliania
  within Haptophyceae.
• We have a few difficulties to analyze the huge
  sequence data perl problem, limited time,
  project organization, etc.