Phylogenetic Assumptions

Transcription

1 Substitution Models and the Phylogenetic Assumptions Vivek Jayaswal Lars S. Jermiin COMMONWEALTH OF AUSTRALIA Copyright htregulation WARNING This material has been reproduced and communicated to you by or on be half of the University of Sydney pursuant to Part VB of the Copyright Act 1968 (the Act). The material in this communication may be subject to copyright under the Act. Any further reproduction or communication of this material by you may be the subject of copyright protection under the Act. Do not remove this notice.

2 Why do we need substitution models? N Human A T C G A C Chimp A G C A A C Gorilla... Orangutan... I1 Root I2 Gorilla Orangutan Chimp Human 4 j= P( u j) P( v j) P( O1 u) P( O2 u) P( O3 v) P( O4 Li = fj u= 1 v= 1 v) I1 P(u j) Root j I2 P(v j) Gorilla Orangutan Chimp Human P(O 1 u) P(O 2 u) P(O 3 3 v) P(O 4 4 u)

3 Topics how correction for multiple substitution are done some of the phylogenetic assumptions how we may evaluate phylogenetic assumptions an example involving bacterial DNA

4 Substitutions at a Single Site Single Substitution Multiple Substitution Coincidental Substitution G A G A A T G T A G A A G A G A A T T Parallel Substitution Convergent Substitution Back Substitution G A G A G G G A G T G A T A A A G A G A A G A Note Every substitution overwrites the evidence of a past state, which leads to an erosion of the historical signal

5 Modeling Nucleotide Substitutions Consider the evolution at a given site in terms of conditional rates-of-change from nucleotide i to nucleotide j A G T C α Aj α AC α AG α AT j A α CA α Cj α CG α CT j C R = α GA α GC α Gj α GT j G α TA α TC α TG α Tj j T Note Here α ij is the conditional rate-of-change from nucleotide i to nucleotide j in R the rate matrix and R is the most general Markov model for DNA

6 Modelling a Site in one Sequence Consider Markov process, X, that results in nucleotide i being converted to nucleotide j over time t P ij [ ] t j ( t)= PXt ()= j X( 0)= i Time If the rates of change are constant, then P(t) = e Rt t = 0 Ancestor X i In matrix notation, this is P ( t )= I + Rt + ( Rt)2 + ( Rt)3 +L 2! 3! ( Rt) k = k =0 k!

7 Modelling a Site in two Sequences Consider two Markov processes, X and Y and the following scenario t F ij [ ( )] Time Y X ()= t PXt ()= i, Y( t)= j X( 0)= Y 0 t = 0 Ancestor In matrix notation, this is ()= P X () t F t ( ) T F 0 ()P Y () t

8 Take-home Message #1 The substitution model, R,, is an integral part of the transition function; it is used to estimate the probability of the present states, given R and t

9 Rate matrix revisited r Aj r AC r AG r AT j A r CA r Cj r CG r CT j C R = i r GA r GC r Gj r GT j G r TA r TC r TG r Tj j T Typically simplified forms of this rate matrix are used These matrices belong to the GTR-family of models and can be represented as r R = r r AC AG AT r r r AC CG CT r r AG CG r GT r r r AT CT GT π A π C π G πt

10 Commonly-used Markov Models Jukes & Cantor (1969) Assumptions 3α α α α α 3α α α R = 1. One rate (α) α α 3α α 2. Uniform nucleotide content α α α 3α Kimura (1980) (2β + α) β α β β (2β + α) β α R = α β (2β + α) β β α β (2β ( β + α) ) 1. Two rates (α and β) 2. Uniform nucleotide content Hasegawa, Kishino & Yano (1985) (βπ Y + απ G ) βπ C απ G βπ T βπ A (βπ R + απ T ) βπ G απ T R = απ A βπ C (βπ Y + απ A ) βπ T βπ A απ C βπ G (βπ R + απ C ) 1. Two rates (α and β) 2. Non-uniform nucleotide content

11 The Phylogenetic Assumptions Given an alignment of nucleotides, phylogenetic methods commonly assume that the sites have evolved under stationary and reversible conditions homogeneous conditions independent and identical conditions

12 Phylogenetic Assumptions Consider the following scenario 1 2 F 1 = [ f A f C f G f T ] F 2 = [ f A f C f G f T ] α Aj α AC α AG α AT j A α CA α Cj α CG α CT j C R 1 = α α α α GA GC Gj GT j G α TA α TC α TG α Tj j T 0 α Aj α AC α AG α AT j A α CA α Cj α CG α CT j C R 2 = α GA α GC α Gj α GT j G α TA α TC α TG α Tj j T [ ] [ ] Π 1 = [ π A π C π G π T ] F 0 = f A f C f G f T Π 2 = π A π C π G π T

13 The Stationary Condition 1 2 Π 1 = π A π C π G π T 0 [ ] Π 2 = [ π A π C π G π T ] F 0 = [ f A f C f G f T ] Note The stationary condition is met if F 0 = Π 1 = Π 2, implying that the marginal distributions of R 1 and R 2 are the same, even though R 1 and R 2 may differ!

14 The Reversible Condition R 1 Π 1 R 2 Π 2 Notes If the process is stationary, then Π 1 = Π 2 Moreover, if π i R ij = π j R ji for all i and j, then the process is reversible

15 The Homogeneous Condition R 1 R 2 Notes The homogeneous condition is met for the Markov processes, R 1 and R 2, if R 1 = R 2 If the homogeneous condition is met, then Π 1 = Π 2 however, non-stationary, and therefore non-reversible, conditions may still prevail (i.e., if F 0 Π 1 = Π 2 )

16 IID condition t 2 1 time t = 0 Y Ancestor X Seq_1 ACGTGTCCATGATTA... R x R x Seq_2 ACCTGCCCAAGATAA... Notes For computational reasons, it is convenient to assume that Sites in the sequence have evolved independently Sites in the sequence have evolved under the same model (R x = R y )

17 Rate Heterogeneity Across Sites RNA-coding Genes

18 Take-home Message #2 Phylogenetic analyses require the users to make certain assumptions about the data before these are investigated in detail

19 Questions Are these phylogenetic assumptions realistic? How can we assess whether the phylogenetic assumptions are met by the data?

20 Answer Inspecting the data before or after inferring the phylogeny increases the chance of finding out what might have taken place during the evolution

21 Considering the IID Condition Visual inspection of the alignment might show whether some regions evolved faster Assume rate-heterogeneity across some sites, and then use phylogenetic methods that account for this

22 Hierarchical Likelihood-Ratio Test Consider the following decision tree Yes No Uniform nucleotide content? No Yes Yes No One conditional rate? No Yes No Yes Two or six conditional rates? Source: Posada & Crandall (1998) Bioinformatics 14,

23 Take-home Message #3 The likelihood-ratio test can be used identify a suitable substitution model for a given data set

24 Testing assumptions prior to modelling n F(t) expected divergence matrix N(t) observed divergence matrix Examples: N(0) = N(t) =

25 Matched-pairs Tests of Symmetry Seq 1 AGACTAGGTCTTGTATAGACTAATGTTCACAGTTTTTTAACTTTGTCAATGGA... Seq 2 AGACGAGGTCGTGTATGGCCTCGTGAGCACGGGTTGTTCACTCCGCCAACGGT... A C G T Σ 2 A C G T Test of Symmetry 2 X Bowker = i< j ( x 2 ij x ) ji x ij + x ji Σ Test of Marginal Symmetry = DV 1 D T, Note: These tests statistics are asymptotically X Stuart D ij = x i x i, χ 2-distributed on ν degrees of freedom V = covariance matrix of D Sources: Bowker (1948). JASA 43, ; Stuart (1955). Biometrica 42,

26 Bacterial 16S rdna Sequences Ribosomal RNA from five bacteria was compared using the matched-pairs test of homogeneity Probabilities Thermotoga Bacillus Deinococcus Thermus Aquifex Thermotoga Bacillus Deinococcus Note It is highly unlike that these data have evolved under homogeneous conditions, implying that it would be unwise to use a time-reversible Markov model Source: Ababneh et al. (2006). Bioinformatics 22,

27 Phylogeny of Bacterial Ribosomal RNA Markov model: GTR Markov model: General Source: Ababneh et al. (2006). Bioinformatics 22,

28 Take-home Message #4 It is important to consider the substitution models carefully when using them in phylogenetic studies

29 Suggested Literature RDM Page, EC Holmes (1998), Molecular Evolution. Chapter 5 (Sections 5.2, 5.3) important reading W-H Li (1997), Molecular Evolution. Chapter 3 (pp ) contains descriptions that are better than those in Page & Holmes 1998 important reading D Posada, KA Crandall, MODELTEST: testing the model of DNA substitution. Bioinformaticsi 14, useful reading LS Jermiin et al. (2008). Phylogenetic model Evaluation. Pp In Bioinformatics - Volume I: Data, Sequences Analysis and Evolution (Ed. Keith J), Humana Press, Totowa, NJ. [2008] important reading