1. Generalizing phylogenetics to
infer patterns predicted by
processes of diversification
Jamie Oaks
Auburn University
phyletica.org
@jamoaks
Scan for slides:
www.slideshare.net/jamieoaks7 © 2007 Boris Kulikov boris-kulikov.blogspot.com

2. © 2007 Boris Kulikov boris-kulikov.blogspot.com

3. © 2007 Boris Kulikov boris-kulikov.blogspot.com

4. © 2007 Boris Kulikov boris-kulikov.blogspot.com
Generalizing phylogenetics to
infer patterns of shared
evolutionary events

5. Phyletica Lab
Postdocs
▶ Perry Wood, Jr
▶ Brian Folt
▶ Jesse Grismer
Graduate students
▶ Tashitso Anamza
▶ Matt Buehler
▶ Kerry Cobb
▶ Kyle David
▶ Saman Jahangiri
▶ Randy Klabacka
▶ Morgan Muell
▶ Tanner Myers
▶ Claire Tracy
▶ Breanna Sipley
▶ Aundrea Westfall
The Phyleticians
Undergraduate students
▶ Laura Lewis
▶ Mary Wells
▶ Hailey Whitaker
▶ Noah Yawn
▶ Charlotte Benedict
▶ Eric Carbo
▶ Ryan Cook
▶ Andrew DeSana
▶ Miles Horne
▶ Jacob Landrum
▶ Nadia L’Bahy
▶ Jorge Lopez-Perez
▶ Holden Smith
▶ Virginia White
▶ Kayla Wilson

6. ▶ Phylogenetics is rapidly becoming the
statistical foundation of biology
© 2007 Boris Kulikov boris-kulikov.blogspot.com

7. ▶ Phylogenetics is rapidly becoming the
statistical foundation of biology
▶ “Big data” present exciting possibilities and
challenges
© 2007 Boris Kulikov boris-kulikov.blogspot.com

8. ▶ Phylogenetics is rapidly becoming the
statistical foundation of biology
▶ “Big data” present exciting possibilities and
challenges
▶ Many opportunities to develop new ways to
study biology in light of phylogeny
© 2007 Boris Kulikov boris-kulikov.blogspot.com

9. phylopic.org CC0

10. ▶ Assumption: All processes of
diversification affect each lineage
independently
phylopic.org CC0

13. Biogeography
▶ Environmental changes that affect whole
communities of species

14. Biogeography
▶ Environmental changes that affect whole
communities of species
Genome evolution
▶ Duplication of a chromosome segment
harboring gene families

15. Biogeography
▶ Environmental changes that affect whole
communities of species
Genome evolution
▶ Duplication of a chromosome segment
harboring gene families
Epidemiology
▶ Transmission at social gatherings

16. Biogeography
▶ Environmental changes that affect whole
communities of species
Genome evolution
▶ Duplication of a chromosome segment
harboring gene families
Epidemiology
▶ Transmission at social gatherings
Endosymbiont evolution (e.g., parasites,
microbiome)
▶ Speciation of the host
▶ Co-colonization of new host species

17. Why account for shared divergences?
J. R. Oaks et al. (2022). PNAS 119: e2121036119

18. Why account for shared divergences?
1. Improve inference
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

19. Why account for shared divergences?
1. Improve inference
2. Provide a framework for
studying processes of
co-diversification
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

20. Biogeography
▶ Environmental changes that affect whole
communities of species
Genome evolution
▶ Duplication of a chromosome segment
harboring gene families
Epidemiology
▶ Transmission at social gatherings
Endosymbiont evolution (e.g., parasites,
microbiome)
▶ Speciation of the host
▶ Co-colonization of new host species

21. Approaches to the problem
A pairwise approach (keep it “simple”)
A fully phylogenetic approach
Tashitso Anamza Tanner Myers
Randy Klabacka Perry Wood, Jr.
Claire Tracy Kerry Cobb
Matt Buehler Nadia L’bahy

22. Approaches to the problem
A pairwise approach (keep it “simple”)
A fully phylogenetic approach
Dr. Perry Wood, Jr.

23. Challenges to accounting for shared divergences
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

24. Challenges to accounting for shared divergences
1. Likelihood for genomic data is
tricky
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

25. Challenges to accounting for shared divergences
1. Likelihood for genomic data is
tricky
2. Lots of possible trees of
different dimensions
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

27. A
A
G
G
G
A

28. A
A
G
G
G
A

29. A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent

30. A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent
▶ Coalescent is a stochastic model of shared inheritance (continuous-time Markov chain =
CTMC)

31. A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent
▶ Coalescent is a stochastic model of shared inheritance (continuous-time Markov chain =
CTMC)
▶ Gene tree branching patterns are a function of population size

32. v
u
G
A
A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent
▶ Coalescent is a stochastic model of shared inheritance (continuous-time Markov chain =
CTMC)
▶ Gene tree branching patterns are a function of population size
▶ Conditional on G, model mutation as a CTMC

33. v
u
G
A
A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent
▶ Coalescent is a stochastic model of shared inheritance (continuous-time Markov chain =
CTMC)
▶ Gene tree branching patterns are a function of population size
▶ Conditional on G, model mutation as a CTMC
▶ Genetic characters provide information about G

34. v
u
G
A
A
A
G
G
G
A
▶ Conditional on “population tree” (T), model “gene trees” (G) using coalescent
▶ Coalescent is a stochastic model of shared inheritance (continuous-time Markov chain =
CTMC)
▶ Gene tree branching patterns are a function of population size
▶ Conditional on G, model mutation as a CTMC
▶ Genetic characters provide information about G
▶ G informs T (population sizes, divergence times, and relationships)

35. v
u
G
A
A
A
G
G
G
A

36. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach

37. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach
▶ Calculate p(genetic data | G) × p(G | T)

38. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach
▶ Calculate p(genetic data | G) × p(G | T)
▶ Use numerical integration (MCMC) to co-estimate both

39. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach
▶ Calculate p(genetic data | G) × p(G | T)
▶ Use numerical integration (MCMC) to co-estimate both
▶ But, G and T are highly correlated

40. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach
▶ Calculate p(genetic data | G) × p(G | T)
▶ Use numerical integration (MCMC) to co-estimate both
▶ But, G and T are highly correlated
▶ As the number of loci (gene trees) increases, MCMC falls apart

41. v
u
G
A
A
A
G
G
G
A
▶ “Standard” hierarchical approach
▶ Calculate p(genetic data | G) × p(G | T)
▶ Use numerical integration (MCMC) to co-estimate both
▶ But, G and T are highly correlated
▶ As the number of loci (gene trees) increases, MCMC falls apart
▶ Can we integrate G analytically?

42. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

43. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

44. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
Q =
(1, 0) (1, 1) (2, 0) (2, 1) · · · (n, n)
(1, 0) · · · · ·
(1, 1) · · · · ·
(2, 0) · · · · ·
(2, 1) · · · · ·
.
.
.
(n, n) · · · · ·
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

45. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
Q =
(1, 0) (1, 1) (2, 0) (2, 1) · · · (n, n)
(1, 0) · · · · ·
(1, 1) · · · · ·
(2, 0) · · · · ·
(2, 1) · · · · ·
.
.
.
(n, n) · · · · ·
Q(n,r);(n,r−1) = (n − r + 1)v, mutation,
Q(n,r);(n,r+1) = (r + 1)u, mutation,
Q(n,r);(n−1,r) =
(n−1−r)n
2Ne (u+v)
, coalescence,
Q(n,r);(n−1,r−1) =
(r−1)n
2Ne (u+v)
, coalescence,
Q(n,r);(n,r) = −
(n−1)n
2Ne (u+v)
− (n − r)v − ru.
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

46. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
Q =
(1, 0) (1, 1) (2, 0) (2, 1) · · · (n, n)
(1, 0) · · · · ·
(1, 1) · · · · ·
(2, 0) · · · · ·
(2, 1) · · · · ·
.
.
.
(n, n) · · · · ·
Q(n,r);(n,r−1) = (n − r + 1)v, mutation,
Q(n,r);(n,r+1) = (r + 1)u, mutation,
Q(n,r);(n−1,r) =
(n−1−r)n
2Ne (u+v)
, coalescence,
Q(n,r);(n−1,r−1) =
(r−1)n
2Ne (u+v)
, coalescence,
Q(n,r);(n,r) = −
(n−1)n
2Ne (u+v)
− (n − r)v − ru.
▶ eQt
to keep track of all conditional probabilites along each branch (Carathéodory-Fejér method1
)
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

47. v
u
G
A
A
A
G
G
G
A
(n, r ) = (3, 1)
(n, r ) = (3, 2)
(n, r ) = (1, 0)
(n, r ) = (1, 0)
(n, r ) = (2, 0)
Q =
(1, 0) (1, 1) (2, 0) (2, 1) · · · (n, n)
(1, 0) · · · · ·
(1, 1) · · · · ·
(2, 0) · · · · ·
(2, 1) · · · · ·
.
.
.
(n, n) · · · · ·
Q(n,r);(n,r−1) = (n − r + 1)v, mutation,
Q(n,r);(n,r+1) = (r + 1)u, mutation,
Q(n,r);(n−1,r) =
(n−1−r)n
2Ne (u+v)
, coalescence,
Q(n,r);(n−1,r−1) =
(r−1)n
2Ne (u+v)
, coalescence,
Q(n,r);(n,r) = −
(n−1)n
2Ne (u+v)
− (n − r)v − ru.
▶ eQt
to keep track of all conditional probabilites along each branch (Carathéodory-Fejér method1
)
▶ At root, get likelihood of population tree integrated over all possible gene trees and mutational
histories2
1
T. Schmelzer and L. N. Trefethen (2007). Electronic Transactions on Numerical Analysis 29: 1–18
2
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

48. Challenges to accounting for shared divergences
1. Likelihood for genomic data is
tricky
2. Lots of possible trees of
different dimensions
True history
τ1
τ2
τ3
Current tree model
τ1 τ2 τ3τ4 τ5 τ6 τ7 τ8
J. R. Oaks et al. (2022). PNAS 119: e2121036119

49. Generalizing tree space

50. Generalizing tree space
nτ = 3
A
B
C
D
A
C
B
D
A
D
B
C
A
B
C
D
A
B
D
C
A
C
B
D
A
C
D
B
A
D
B
C
A
D
C
B
B
C
A
D
B
C
D
A
B
D
A
C
B
D
C
A
C
D
A
B
C
D
B
A

51. Generalizing tree space
nτ = 3 nτ = 2 nτ = 1
A
B
C
D
A
C
B
D
A
D
B
C
A
B
C
D
A
C
B
D
A
D
B
C
A
B
C
D
A
B
D
C
A
C
B
D
A
B
C
D
A
C
B
D
A
D
B
C
A
C
D
B
A
D
B
C
A
D
C
B
B
C
A
D
B
D
A
C
C
D
A
B
A
B
C
D
B
C
A
D
B
C
D
A
B
D
A
C
A
B
C
D
A
B
D
C
A
C
D
B
B
D
C
A
C
D
A
B
C
D
B
A
B
C
D
A

52. Generalizing tree space
nτ = 3 nτ = 2 nτ = 1
A
B
C
D
A
C
B
D
A
D
B
C
A
B
C
D
A
C
B
D
A
D
B
C
A
B
C
D
A
B
D
C
A
C
B
D
A
B
C
D
A
C
B
D
A
D
B
C
A
C
D
B
A
D
B
C
A
D
C
B
B
C
A
D
B
D
A
C
C
D
A
B
A
B
C
D
B
C
A
D
B
C
D
A
B
D
A
C
A
B
C
D
A
B
D
C
A
C
D
B
B
D
C
A
C
D
A
B
C
D
B
A
B
C
D
A

53. Generalized tree distribution
▶ All topologies equally probable
▶ Parametric distribution on age
of root
▶ Beta distributions on other
divergence times
τ0
τ1 ∼ Beta(τ0, τ3, α, β)
*
τ2 ∼ Beta(τ0, τ3, α, β)
*
τ3 ∼ Beta(τ0, τ4, α, β)
τ4 ∼ Gamma(k, θ)
*
I
H
G
F
E
D
C
B
A
t6
t3
t1
t2
t4
t5
J. R. Oaks et al. (2022). PNAS 119: e2121036119

54. Inferring trees with shared divergences
Split
Merge
Reversible-jump MCMC
J. R. Oaks et al. (2022). PNAS 119: e2121036119

55. Validating rjMCMC with 7-leaf tree
350 400 450
0.000
0.005
0.010
0.015
0.020
Number of times each topology sampled
Density
Binom(5e+07, 1/127,569)
MCMC
χ2
= 128,046
p = 0.172
The rjMCMC algorithms sample the expected generalized tree distribution
J. R. Oaks et al. (2022). PNAS 119: e2121036119

56. Phylogenetic coevality
Phyco val
J. R. Oaks et al. (2022). PNAS 119: e2121036119
Ecoevolity
Estimating evolutionary coevality
J. R. Oaks (2019). Systematic Biology 68: 371–395
▶ Tree model
▶ rjMCMC sampling of generalized tree distribution
1
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

57. Phylogenetic coevality
Phyco val
J. R. Oaks et al. (2022). PNAS 119: e2121036119
Ecoevolity
Estimating evolutionary coevality
J. R. Oaks (2019). Systematic Biology 68: 371–395
▶ Tree model
▶ rjMCMC sampling of generalized tree distribution
▶ Likelihood model
▶ CTMC model of characters evolving along genealogies
▶ Infer species trees by analytically integrate over genealogies1
1
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

58. Phylogenetic coevality
Phyco val
J. R. Oaks et al. (2022). PNAS 119: e2121036119
Ecoevolity
Estimating evolutionary coevality
J. R. Oaks (2019). Systematic Biology 68: 371–395
▶ Tree model
▶ rjMCMC sampling of generalized tree distribution
▶ Likelihood model
▶ CTMC model of characters evolving along genealogies
▶ Infer species trees by analytically integrate over genealogies1
▶ Goal: Co-estimation of phylogeny and shared divergences from genomic data
1
D. Bryant et al. (2012). Molecular Biology and Evolution 29: 1917–1932

59. Does it work?

60. Methods: Simulations
▶ Simulated 100 data sets with 50,000 base
pairs
τ1
0.05
τ2
0.08
τ3
0.2
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2

61. Methods: Simulations
▶ Simulated 100 data sets with 50,000 base
pairs
▶ Analyzed each data set with:
▶ MG = Generalized tree model
▶ MIB = Independent-bifurcating tree model
τ1
0.05
τ2
0.08
τ3
0.2
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2

62. Methods: Simulations
▶ Simulated 100 data sets with 50,000 base
pairs
▶ Analyzed each data set with:
▶ MG = Generalized tree model
▶ MIB = Independent-bifurcating tree model
▶ Simulated 100 data sets where topology and
div times randomly drawn from MG and MIB
τ1
0.05
τ2
0.08
τ3
0.2
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2

63. 0.05
0.08
0.2
J. R. Oaks et al. (2022). PNAS 119: e2121036119

64. 0.0 0.5 1.0
0.05
0.0 0.5 1.0
0.08
0.2
0.0 0.5 1.0
0.0 0.5 1.0
0.0 0.5 1.0
0.0 0.5 1.0
J. R. Oaks et al. (2022). PNAS 119: e2121036119

65. τ1
0.05
τ2
0.08
τ3
0.2
t5
t2
t1
t3
t4
J. R. Oaks et al. (2022). PNAS 119: e2121036119

66. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.05
τ2
0.08
τ3
0.2
t5
t2
t1
t3
t4
Splitting t4
0.00
0.25
0.50
0.75
1.00
Posterior
probability
J. R. Oaks et al. (2022). PNAS 119: e2121036119

67. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.05
τ2
0.08
τ3
0.2
t5
t2
t1
t3
t4
0.000
0.002
0.004
0.006
0.008
0.010
0.012
p = 4.1 × 10−18
MG MIB
Distance
from
true
tree
J. R. Oaks et al. (2022). PNAS 119: e2121036119

68. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.05
τ2
0.08
τ3
0.2
t5
t2
t1
t3
t4
0.000
0.002
0.004
0.006
0.008
0.010
0.012
p = 4.1 × 10−18
MG MIB
Distance
from
true
tree
MG significantly better at inferring trees with shared divergences
J. R. Oaks et al. (2022). PNAS 119: e2121036119

69. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2
t8
t2
t1
t7
t4
t3
t6
t5
J. R. Oaks et al. (2022). PNAS 119: e2121036119

70. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2
t8
t2
t1
t7
t4
t3
t6
t5
True topology
0.00
0.25
0.50
0.75
1.00
Posterior
probability
J. R. Oaks et al. (2022). PNAS 119: e2121036119

71. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2
t8
t2
t1
t7
t4
t3
t6
t5
0.0000
0.0025
0.0050
0.0075
0.0100
0.0125
p = 0.36
MG MIB
Distance
from
true
tree
J. R. Oaks et al. (2022). PNAS 119: e2121036119

72. MG = Generalized model MIB = Independent-bifurcating model
τ1
0.03
τ2
0.05
τ3
0.06
τ4
0.11
τ5
0.12
τ6
0.14
τ7
0.18
τ8
0.2
t8
t2
t1
t7
t4
t3
t6
t5
0.0000
0.0025
0.0050
0.0075
0.0100
0.0125
p = 0.36
MG MIB
Distance
from
true
tree
MG performs as well as true model when divergences are independent
J. R. Oaks et al. (2022). PNAS 119: e2121036119

73. Results: random MG trees
0.5 1.0 1.5 2.0 2.5
0.5
1.0
1.5
2.0
2.5 p(TL ∈ CI) = 0.96
RMSE = 0.00486
True tree length
Estimated
tree
length
J. R. Oaks et al. (2022). PNAS 119: e2121036119

74. Results: random MG trees
0.5 1.0 1.5 2.0 2.5
0.5
1.0
1.5
2.0
2.5 p(TL ∈ CI) = 0.96
RMSE = 0.00486
True tree length
Estimated
tree
length
Shared divergence
0.00
0.25
0.50
0.75
1.00
Posterior
probability J. R. Oaks et al. (2022). PNAS 119: e2121036119

75. Results: random MG trees
0.5 1.0 1.5 2.0 2.5
0.5
1.0
1.5
2.0
2.5 p(TL ∈ CI) = 0.96
RMSE = 0.00486
True tree length
Estimated
tree
length
Shared divergence
0.00
0.25
0.50
0.75
1.00
Posterior
probability
MG performs well with data simulated on random trees with shared divergences
J. R. Oaks et al. (2022). PNAS 119: e2121036119

76. Results: random MIB trees
Probability of incorrectly merged divergence times (true model = MIB)
Merged times
0.00
0.25
0.50
0.75
1.00
FPR = 0.047
A
Posterior
probability
0.00 0.05 0.10 0.15 0.20
0.00
0.25
0.50
0.75
1.00
B
Time difference
Posterior
probability 0.0 0.1 0.2 0.3
0.00
0.25
0.50
0.75
1.00
C
Midpoint divergence time
Posterior
probability
J. R. Oaks et al. (2022). PNAS 119: e2121036119

77. Results: random MIB trees
Probability of incorrectly merged divergence times (true model = MIB)
Merged times
0.00
0.25
0.50
0.75
1.00
FPR = 0.047
A
Posterior
probability
0.00 0.05 0.10 0.15 0.20
0.00
0.25
0.50
0.75
1.00
B
Time difference
Posterior
probability 0.0 0.1 0.2 0.3
0.00
0.25
0.50
0.75
1.00
C
Midpoint divergence time
Posterior
probability
MG has low false positive rate
J. R. Oaks et al. (2022). PNAS 119: e2121036119

78. MG = Generalized model MIB = Independent-bifurcating model
0.000
0.002
0.004
0.006
0.008
0.010
0.012
Ave
std
dev
of
split
freq
(ASDSF)
Generalizing tree space improves MCMC convergence and mixing
J. R. Oaks et al. (2022). PNAS 119: e2121036119

79. Scan for sea-level animation

80. Scan for sea-level animation

81. Scan for sea-level animation
Did fragmentation of islands
promote diversification?

82. Cyrtodactylus Gekko
©Rafe M. Brown
10
15
20
118 120 122 124 126
Longitude
Latitude
10
15
20
118 120 122 124 126
Longitude
Latitude
©Rafe M. Brown
J. R. Oaks et al. (2022). PNAS 119: e2121036119

83. Cyrtodactylus Gekko
©Rafe M. Brown
10
15
20
118 120 122 124 126
Longitude
Latitude
10
15
20
118 120 122 124 126
Longitude
Latitude
©Rafe M. Brown
1702 loci
155,887 sites
1033 loci
94,813 sites
J. R. Oaks et al. (2022). PNAS 119: e2121036119

84. Gekko
5°N
10°N
15°N
20°N
116°E 118°E 120°E 122°E 124°E 126°E
Longitude
Latitude
D
116°E 118°E 120°E 122°E 124°E 126°E
Longitude
0.28
0.55
0.19
0.35
0.81
1.0
n
c
i
l
P s
l
P
30 20 10 0
1
2
14
15
16
17
19
18
20
22
21
23
24
26
25
4
3
5
7
6
13
12
11
10
9
8
0.87
0.86
n
c
i
l
P s
l
P
30 20 10 0
C
G. athymus
G. carusadensis
G. crombota
G. ernstkelleri
G. gigante
G. mindorensis
G. monarchus
G. porosus
G. romblon
G. rossi
G. sp. a
G. sp. b
1
5
11
2
7
6
22
20
16
19
18
15
23
17
21
26
25
24
14
13
12
4
3
10
9
8
Oligocene Miocene
Oligocene Miocene
J. R. Oaks et al. (2022). PNAS 119: e2121036119

85. Cyrtodactylus
D
5°N
10°N
15°N
20°N
116°E 118°E 120°E 122°E 124°E 126°E
Longitude
Latitude
B
1.0
1
10
24
23
25
27
26
18
19
22
21
20
14
13
15
17
16
12
11
7
9
8
2
3
4
6
5
0.71
0.75
n
c
i
l
P s
l
P
30 20 10 0
A
C
C. annulatus
C. gubaot
C. jambangan
C. mamanwa
C. philippinicus
C. redimiculus
C. sumuroi
C. tautbatorum
0.30
0.45
0.90
0.89
0.57
10
13
17
6
4
5
9
3
7
24
27
18
19
22
23
14
26
21
25
15
16
20
12
11
1
8
2
Oligocene Miocene
J. R. Oaks et al. (2022). PNAS 119: e2121036119

86. Take-home points
▶ We can accurately infer phylogenies with shared divergences

87. Take-home points
▶ We can accurately infer phylogenies with shared divergences
▶ Generalizing tree space avoids spurious support incorrect relationships and improves
MCMC mixing

88. Take-home points
▶ We can accurately infer phylogenies with shared divergences
▶ Generalizing tree space avoids spurious support incorrect relationships and improves
MCMC mixing
▶ Among Philippine gekkonids, we found support for shared divergences predicted by
sea-level changes

89. Open science: everything is available. . .
Software:
▶ Phycoeval: github.com/phyletica/ecoevolity
(release coming soon)
Open-Science Notebooks:
▶ Phycoeval analyses: github.com/phyletica/phycoeval-experiments
▶ Gecko RADseq: github.com/phyletica/gekgo

90. Moving forward: Theory/methods

91. Moving forward: Theory/methods
▶ Develop process-based and
trait-dependent distributions over the
space of generalized trees
▶ “Birth-death-burst” model

92. Moving forward: Theory/methods
▶ Develop process-based and
trait-dependent distributions over the
space of generalized trees
▶ “Birth-death-burst” model
▶ Extend generalized tree distribution to
trees that are not ultrametric

93. Moving forward: Theory/methods
▶ Develop process-based and
trait-dependent distributions over the
space of generalized trees
▶ “Birth-death-burst” model
▶ Extend generalized tree distribution to
trees that are not ultrametric
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

94. Moving forward: Theory/methods
▶ Develop process-based and
trait-dependent distributions over the
space of generalized trees
▶ “Birth-death-burst” model
▶ Extend generalized tree distribution to
trees that are not ultrametric
▶ Couple generalized tree distribution with
other phylogenetic likelihood models
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

95. Moving forward: Applications
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

96. Moving forward: Applications
Epidemiological dynamics of “super-spreading”
events during the COVID-19 pandemic
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

97. Moving forward: Applications
Epidemiological dynamics of “super-spreading”
events during the COVID-19 pandemic
▶ Spread at social gatherings creates shared
and multifurcating divergences in the viral
“transmission tree”
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

98. Moving forward: Applications
Epidemiological dynamics of “super-spreading”
events during the COVID-19 pandemic
▶ Spread at social gatherings creates shared
and multifurcating divergences in the viral
“transmission tree”
▶ Estimate rate of shared divergences as
proxy for spread via social gatherings
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

99. Moving forward: Applications
Epidemiological dynamics of “super-spreading”
events during the COVID-19 pandemic
▶ Spread at social gatherings creates shared
and multifurcating divergences in the viral
“transmission tree”
▶ Estimate rate of shared divergences as
proxy for spread via social gatherings
▶ Test if this varies over time, among
regions, and among variants of
SARS-CoV-2
nextstrain.org J. Hadfield et al. (2018). Bioinformatics 34: 4121–4123

100. © 2007 Boris Kulikov boris-kulikov.blogspot.com

102. Come work with us!
Adaptive Comp Bio Internships!
Scan to internship listing:
www.adaptivebiotech.com/career-listings

103. Thanks everyone!
▶ Thanks Devang and Duke GCB & CBB!
▶ Bryan Howie and my team at Adaptive
▶ Phyletica Lab (the Phyleticians)
▶ Mark Holder
▶ Rafe Brown
▶ Cam Siler
▶ Lee Grismer
Computation:
▶ Alabama Supercomputer Authority
▶ Auburn University Hopper Cluster
Funding:
Photo credits:
▶ Rafe Brown
▶ Perry Wood, Jr.
▶ PhyloPic

104. Questions?
joaks@auburn.edu
@jamoaks
phyletica.org
Scan for slides:
www.slideshare.net/jamieoaks7
© 2007 Boris Kulikov boris-kulikov.blogspot.com