Introduction to phylogenetic/phylogenomic concepts and methods

1
Introduction to phylogenetic/phylogenomic
concepts and methods

• With a brief discussion of phylogenomic methods
  of species phylogeny estimation and orthology
  prediction methods

2

• "Nothing in Biology Makes Sense Except in the
  Light of Evolution"
• Theodosius Dobzhansky (1973)

3
Revised tree of life including horizontal transfer
Horizontal transfer endosymbiosis
4
Is Phylogenetic Analysis Truth?

• No -- its just a prediction, and like any other
  prediction method, it is prone to errors of
  different types.
• Its important that you be aware of sources of
  possible errors in a phylogenetic tree
  reconstruction
• However, phylogenetic reconstruction is one of
  the best tools available for understanding both
  how species evolve and how gene families evolve
  novel functions and structures.
• As Winston Churchill said

5
Winston Churchill
http//www.winstonchurchill.org/
6
Workflow for phylogenomic analysis
Sjölander, Phylogenomic inference of protein
molecular function advances and challenges
Bioinformatics (2004) 20 (2) 170-179.
7
Delsuc et al, Phylogenomics and the
reconstruction of the tree of life Nature
Reviews Genetics 6, 361-375 (May 2005)
8
Trees are a special type of graph

• Graphs have nodes (vertices) and edges (branches)
• Edges can be directed or undirected
• Nodes can be internal or terminal
• Terminal nodes in a phylogenetic tree are called
  leaves (or taxa)
• The term taxon refers to (groups of) species,
  but is commonly used to describe genes in
  multi-gene families, even when the same species
  may be found in multiple copies in the tree
• Trees are a special subtype of graph (acyclic
  connected graphs)
• The valency (or degree) of a node equals the
  number of edges
• A tree for which every internal node (except for
  the root) has degree 3 (one ancestor and two
  children) is called a bifurcating or binary
  tree.
• Trees for which internal nodes can have gt2
  children are called multifurcating trees
• The diameter of a tree is equal to the longest
  path between two leaves (including edge lengths,
  not simply number of edges)
• Most phylogenetic trees are unrooted, and special
  methods must be used to infer the root.

9
Interpreting tree topologies

• Many phylogenetic trees are not meant to be
  interpreted as rooted (more about this later)
• Terminal nodes (leaves) represent contemporary
  taxa (organisms, genes, proteins, or other
  objects)
• Internal nodes represent inferred ancestors
• Edge lengths are supposed to be proportional to
  the evolutionary distance

10
node
human
A clade
mouse
Fruit fly
Root?

Taxa (singular taxon)
Terminal nodes (leaves)
From Bioinformatics, A practical guide to the
analysis of genes and proteins Edited by
Baxevanis Ouellette
11
Finding an optimal tree topology is unlikely!

• Number of (unrooted) binary trees on n leaves is
  (2n-5)!!
• If each tree on 1000 taxa could be analyzed in
  0.001 seconds, we would find the best tree in
• 2890 millennia

12
Morphological vs. Molecular

• Classical phylogenetic analysis morphological
  features
• Presence of vertebra, number of legs, fur vs
  scales, etc.
• Molecular evolution Using molecular features
  (DNA, RNA and proteins)
• Analysis based on homologous sequences (e.g.,
  globins) in different species
• Additional issues
• restriction to orthologs is critical when
  reconstructing species phylogenies.
• If estimating a gene family phylogeny, youre
  better off gathering all homologs and figuring
  out orthologs from the tree topology.
• Slow-evolving genes are best for reconstructing
  phylogenies of distant species use protein
  sequences
• Fast-evolving genes (and use of nucleotide data)
  may be more effective at reconstructing
  phylogenies of closely related species

13
Rooted vs unrooted trees

• Most trees are unrooted -- you need to take
  additional actions to root the tree
• Rooting a species tree not so hard
• Rooting a protein superfamily tree (with gene
  duplication) hard

14
Two main classes of tree inference methods

• Distance-based
• Input is a matrix of distances between species
• Can be fraction of residues they disagree on, or
  -alignment score between them, or
• E.g., Neighbor-Joining, UPGMA, etc.
• Character-based
• Examine each character (e.g., residue) separately
• E.g., Maximum Likelihood, Maximum Parsimony, Mr.
  Bayes
• Question to test understanding
• Is SATCHMO a distance-based or character-based
  method? What about SCI-PHY?

15
Distance-based Phylogenetic Methods
T. Warnow
16
Maximum Parsimony

• Input Set S of n aligned sequences of length k
• Output
• A phylogenetic tree T leaf-labeled by sequences
  in S
• additional sequences of length k labeling the
  internal nodes of T
• such that is minimized.

E(T) is the set of edges in the tree T H is the
Hamming distance (minimum substitutions) between
edges (i,j).
T. Warnow
17
The fundamental principle behind maximum parsimony

• Occams Razor
• Entia non sunt multiplicanda praeter
  necessitatem.
• William of Occam (1300-1349)

The best tree is the one which requires the least
number of substitutions
18
Maximum Likelihood

• Input Set S of n aligned sequences of length k,
  and a specified parametric model
• Output
• A phylogenetic tree T leaf-labeled by sequences
  in S
• With additional model parameters (e.g. edge
  lengths)
• such that PrS(T, parameters) is maximized.
• (Recall that ML methods seek to identify a model
  that maximizes the probability of the data.)

19
Maximum Likelihood (in English)

• Require a model of evolution
• Each substitution has an associated likelihood
  given a branch of a certain length
• A function is derived to represent the likelihood
  of the data given the tree, branch-lengths and
  additional parameters
• Find the tree that maximizes the likelihood

20
Molecular Clock

• UPGMA implicitly assumes that all distances are
  clock-like

2
3
2
3
4
1
1
4
21
More about molecular clocks

• The molecular clock (based on the molecular clock
  hypothesis (MCH)) is a technique in genetics to
  date when two species diverged. Elapsed time is
  deduced by applying a time scale to the number of
  molecular differences measured between the
  species' DNA sequences or proteins.

http//en.wikipedia.org/wiki/Molecular_clock
22
More about molecular clocks

• The notion of the existence of a so-called
  "molecular clock" was first attributed to Emile
  Zuckerkandl and Linus Pauling who, in 1962,
  noticed that the number of amino acid differences
  in hemoglobin between lineages scales roughly
  with divergence times, as estimated from fossil
  evidence. They generalized this observation to
  assert that the rate of evolutionary change of
  any specified protein was approximately constant
  over time and over different lineages.
• Later Allan Wilson and Vincent Sarich built upon
  this work and the work of Motoo Kimura observed
  and formalized that rare spontaneous errors in
  DNA replication cause the mutations that drive
  molecular evolution, and that the accumulation of
  evolutionarily "neutral" differences between two
  sequences could be used to measure time, if the
  error rate of DNA replication could be calibrated.

http//en.wikipedia.org/wiki/Molecular_clock
23
Quantifying Error
FN
FN false negative (missing edge) FP false
positive (incorrect edge)
FP
Courtesy of T. Warnow
24
Two masking approaches

• Construct many MSAs for same set of sequences
  (varying alignment parameters), concatenate
  alignments, and estimate trees (W. Wheeler)
• Expectation is that uncertain regions of
  alignment will cancel each other out, whereas
  signal from regions with consensus support will
  be magnified
• Delete noisy columns
• Options include
• Remove gappy columns only
• Remove gappy and noisy columns

25
Rooting the tree (or not)

• Most tree methods produce unrooted trees (even if
  they appear to be rooted)
• In construction of species trees, use of an
  outgroup sequence is recommended (and relatively
  straightforward)
• Choosing an outgroup for protein superfamily
  reconstruction is not so straightforward
• Alternatives
• Root the tree by finding the midpoint on the
  longest path between any pair of leaves (assumes
  molecular clock)
• Root the tree by locating the root that requires
  the minimal number of mutations from the ancestor
  to current sequences

26
Rooting the tree using an outgroup sequence

• Choosing an outgroup
• For species trees choose a related organism that
  is more distant from any of the ingroup sequences
  than they are to each other.
• For protein family trees choose a related
  protein that is evolutionarily more distant from
  any ingroup sequence than any are to each other
• (how to do this since we dont know the
  evolutionary history??)
• Place the root somewhere between the outgroup and
  the rest (either on the node or in a branch)

27
Bootstrap Analysis

• Columns from the input alignment are resampled
  with replacement to create many bootstrap
  replicate data sets
• each resampled MSA will have the same number of
  columns -- some columns may be sampled multiple
  times, and others not at all
• A phylogenetic tree is constructed for each
  bootstrap replicate data set (e.g. with
  parsimony, distance, ML etc.)
• Agreement among the resulting trees is
  summarized with a majority-rule consensus tree
  (e.g., PHYLIP consense program)
• Frequencies of occurrence of groups, bootstrap
  proportions (BPs), are a measure of support for
  those groups
• Bootstrap analysis does not actually tell you the
  expected accuracy of that subtree -- just the
  support for that subtree in the alignment

28
Phylogenetic uncertainty and subtree neighbors
If no known function(s) are available for a group
of sequences (e.g., subtree A, in the figure
above), function may be inferred based on data
available for sequences in a neighboring subtree.
(This idea is referred to as subtree neighbors
by Zmasek and Eddy.) However, the coarse
branching order between clades can be highly
variable. If you build four trees (NJ, ML, MP, Mr
Bayes) you may find four (or more!) different
tree topologies. Example SH2 domains. SH2
domains bind phosphopeptides the residues
responsible for binding are generally conserved
within subfamilies but vary across subfamilies.
In cases where we know something about the
function of members of a family, can we use this
information to evaluate which of the different
trees might be more reliable?
Src homology 2 domain 1SPSA
29
Recommendations on phylogeny estimation

• Most important factor is the quality of the input
  data
• Look at the data from as many angles as possible.
  Does this sound familiar??
• Choice of outgroup taxa can have as much
  influence as the ingroup taxa.
• Compute every analysis with several outgroups?
• Recognize that different programs can be
  influenced by the order of the analysis, i.e. the
  order in which the sequences appear in your file.
  
• PHYLIP and PAUP offer a "jumble" option that
  reruns the analysis with different input orders.

http//biology.unm.edu/biology/maggieww/Public_Htm
l/phylogeny.htm
30
Orthology prediction
31
Major papers on orthology prediction

• Fitch, W.M. (2000) Homology a personal view on
  some of the problems. Trends Genet, 16, 227-231.
• Sonnhammer and Koonin, Orthology, paralogy and
  proposed classification for paralog subtypes
  Trends Genetics 2002 (Introduces key terms such
  as inparalog)
• Remm, M., Storm, C.E. and Sonnhammer, E.L. (2001)
  Automatic clustering of orthologs and in-paralogs
  from pairwise species comparisons. J Mol Biol,
  314, 1041-1052. (InParanoid method)
• Christian M Zmasek and Sean R Eddy (2002) RIO
  Analyzing proteomes by automated phylogenomics
  using resampled inference of orthologs, BMC
  Bioinformatics (introduces the terms
  super-ortholog and ultra-paralog)
• Tatusov, R.L. et al. (2003) The COG database an
  updated version includes eukaryotes. BMC
  Bioinformatics, 4, 41.
• Chen, F., Mackey, A.J., Stoeckert, C.J., Jr. and
  Roos, D.S. (2006) OrthoMCL-DB querying a
  comprehensive multi-species collection of
  ortholog groups. Nucleic Acids Res, 34, D363-368.
• Jensen, L.J., Julien, P., Kuhn, M., von Mering,
  C., Muller, J., Doerks, T. and Bork, P. (2008)
  eggNOG automated construction and annotation of
  orthologous groups of genes. Nucleic Acids Res,
  36, D250-254.
• Ruan, J., Li, H., Chen, Z., Coghlan, A., Coin,
  L.J., Guo, Y., Heriche, J.K., Hu, Y.,
  Kristiansen, K., Li, R. et al. (2008) TreeFam
  2008 Update. Nucleic Acids Res, 36, D735-740.
  (phylogenomic identification of orthologs, but
  restricted to animal genomes)
• Ruchira S. Datta, Christopher Meacham, Bushra
  Samad, Christoph Neyer and Kimmen Sjölander
  (2009) "Berkeley PHOG PhyloFacts Orthology Group
  Prediction Web Server," Nucleic Acids Research
  (2009 Web server issue)

32
Other papers about orthologs

• Greg Petsko "Homologuephobia", Genome Biology
  2001
• Eugene Koonin, "An Apology for Orthologs - or
  brave new memes", Genome Biology 2001
• R A Jensen "Orthologs and Paralogs - we need to
  get it right", Genome Biology 2001 (includes a
  discussion of conflicting opinions by Eugene
  Koonin and Greg Petsko on this subject)
• Iddo Friedberg, "Automated Function Prediction
  the Genomic Challenge", Briefings in
  Bioinformatics, 2006 (includes a discussion of
  how function prediction methods are evaluated)

33
Why is orthology important?
34
Orthology definitions
Orthology (standard definition) the MRCA must
correspond to a speciation event. (By this
definition, the yeast sequence is orthologous to
both H1 and H2, which are co-orthologs to the
yeast sequence.) Super-orthology is more
restrictive than orthology all nodes on a path
between two leaves must correspond to a
speciation event. (Zmasek Eddy, 2002)
S
Super-orthologs
D
H1 C1 M1 R1 F1 W1 H2 C2 M2 R2
F2 W2
Yeast
Advantages of super-orthology it is transitive,
and provides a secure basis for inference of
function. Disadvantages highly restrictive (low
coverage)
Human, Chimp, Mouse, Rat, Fly, Worm
35
New terms

• Co-ortholog
• Inparalog
• Outparalog

Sonnhammer and Koonin, Orthology, paralogy and
proposed classification for paralog subtypes,
Trends in Genetics, 2002
36
Orthologs and paralogs - we need to get it right,
Roy A Jensen, Genome Biology 2001
37
Gene tree-species tree reconciliation is used to
identify orthologs
38
Ortholog prediction accuracy
PHOG-O Standard orthology definition PHOG-S
Super-orthologs PHOG-T thresholded
PHOGs PHOG-T(M) optimized for mouse PHOG-T(Z)
optimized for zebrafish PHOG-T(F) optimized for
fruit fly
Benchmark dataset 100 (non-homologous) human
sequences from TreeFam-A, filtered to remove
homologs, and manually curated orthologs from
mouse, zebrafish and fruit fly.
39
PHOG search results genes can belong to
multiple orthology groupsbased on regions
selected, or time points at which sequences were
clustered
40
How is PHOG different from InParanoid, OrthoMCL
and TreeFam?

• PHOG is based on precalculated trees in
  PhyloFacts 55K protein families (and expanding
  daily)
• PhyloFacts includes trees for individual domains
  as well as entire domain architectures
• PhyloFacts trees include sequences from across
  the Tree of Life (all species, not restricted to
  whole genomes)
• TreeFam is restricted to animal genomes
• TreeFam uses gene tree-species tree
  reconciliation (PHOG does not)
• This should enable TreeFam to be more accurate,
  assuming the species tree is correct (note this
  cannot always be assumed)
• Advantages of PHOG
• Individual domains provide different evolutionary
  perspectives
• Improved taxonomic sampling (known to be critical
  for tree topology accuracy)
• Requiring global homology is often too
  restrictive and can cause topology errors
• Allowing local matches avoids problems with
  rejecting sequences due to gene model errors
  (very common among eukaryotes)

41
Relevant questions

• What are the uses of orthologs?
• What is the (accepted/common) definition of
  ortholog and paralog?
• What is the most common method used to identify
  orthologs? Is this approach rigorous?
• How can you be sure you have an ortholog? (What
  evidence is required to assert two proteins are
  orthologs?)
• What is the actual significance of two proteins
  being orthologous?
• How is orthology accuracy assessed? Is there a
  standard?
• What is the expected accuracy of ortholog
  identification using different methods?