ABC SMC for parameter estimation and model selection with - - PowerPoint PPT Presentation

▶
abc smc for parameter estimation and model selection with
SMART_READER_LITE
LIVE PREVIEW

ABC SMC for parameter estimation and model selection with - - PowerPoint PPT Presentation

May 07, 2023 30 likes •580 views

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011 ABC SMC for parameter estimation and model selection with applications in systems biology Tina Toni Department of Biological Engineering, Massachusetts Institute of

slide-1
SLIDE 1

ABC SMC for parameter estimation and model selection with applications in systems biology

Tina Toni

Department of Biological Engineering, Massachusetts Institute of Technology, USA & Theoretical Systems Biology Group, Imperial College London, UK

ABC in London, 05/05/2011

Tina Toni ABC in systems biology 05/05/2011 1 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-2
SLIDE 2

Motivation

Complex biological systems

Models often ODE or stochastic master equations High dimensional parameter space Time course, non-equidistant, missing data

Interested in

Characterization of distributions over parameters rather than point estimates. What dynamic behaviour can reproduce data? Which models represent suitable hypothesis about the system?

Tina Toni ABC in systems biology 05/05/2011 2 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-3
SLIDE 3

Outline

1

Parameter estimation ABC basics ABC SMC Application: Modeling bacterial stress response

2

Model selection ABC SMC for model selection Application: Epo signaling pathway Application: Phosphorylation dynamics

Tina Toni ABC in systems biology 05/05/2011 3 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-4
SLIDE 4

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-5
SLIDE 5

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-6
SLIDE 6

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-7
SLIDE 7

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-8
SLIDE 8

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-9
SLIDE 9

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Tina Toni ABC in systems biology 05/05/2011 4 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-10
SLIDE 10

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Tina Toni ABC in systems biology 05/05/2011 4 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-11
SLIDE 11

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-12
SLIDE 12

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-13
SLIDE 13

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D) P(θ|dist(D, Dc) ≤ ǫ)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-14
SLIDE 14

Parameter estimation ABC basics

Approximate Bayesian Computation basics

1 Sample θc from P(θ). 2 Simulate a data set Dc from the model with θc. 3 If dist(D, Dc) ≤ ǫ, accept θc, otherwise reject. 4 Return to 1. Prior Posterior P(θ) P(θ|D) P(θ|dist(D, Dc) ≤ ǫ)

x x x x time

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-15
SLIDE 15

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Prior Posterior

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-16
SLIDE 16

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-17
SLIDE 17

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-18
SLIDE 18

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-19
SLIDE 19

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-20
SLIDE 20

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-21
SLIDE 21

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-22
SLIDE 22

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1 Population 2

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-23
SLIDE 23

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1 Population 2 Population T

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-24
SLIDE 24

Parameter estimation ABC SMC

ABC SMC (Sequential Monte Carlo)

Intermediate Distributions Prior Posterior ǫ1 ǫ2 . . . ǫT

Population 1 Population 2 Population T

(Sisson et al., 2007, PNAS)

Tina Toni ABC in systems biology 05/05/2011 5 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-25
SLIDE 25

Parameter estimation ABC SMC

Weights

wt(θt) = πt(θt) ηt(θt) ηt(θt) = 1 (π(θt) > 0) 1 (dist < ǫt)

  • πt−1(θt−1)Kt(θt|θt−1)dθt−1

w(i)

t

= π(θ(i)

t )

N

j=1 w(j) t−1Kt(θ(i) t |θ(j) t−1)

(Toni et al., J. R. Soc. Interface, 2009)

Tina Toni ABC in systems biology 05/05/2011 6 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-26
SLIDE 26

Parameter estimation Application: Modeling bacterial stress response

Application: Modeling phage shock protein response in Escherichia coli

Signal

Phage damages the membrane of bacteria (we call this stress).

Response

Reduced motility Anaerobic respiration Start membrane repair mechanisms etc.

Tina Toni ABC in systems biology 05/05/2011 7 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-27
SLIDE 27

Parameter estimation Application: Modeling bacterial stress response

Phage shock protein response

H+

H+

H+

H+ Normal conditions Normal conditions ArcB PspA B C PspF Stress conditions ArcB PspA B C PspF

Psp response

1

Stress

2

PspC changes conformation

3

PspB changes conformation

4

Complex PspA-PspF breaks

5

PspF is free to act as a transcription factor

6

Psp genes (A,B,C,D,E,F,G) → proteins

7

Response

Tina Toni ABC in systems biology 05/05/2011 8 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-28
SLIDE 28

Parameter estimation Application: Modeling bacterial stress response

Data and Questions of interest

Data

Details of molecular interactions. Kinetic parameters unknown. Few data available (qualitative end point data).

Tina Toni ABC in systems biology 05/05/2011 9 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-29
SLIDE 29

Parameter estimation Application: Modeling bacterial stress response

Data and Questions of interest

Data

Details of molecular interactions. Kinetic parameters unknown. Few data available (qualitative end point data).

Questions of interest

1 What dynamic behaviour is possible? 2 Can it be inferred from qualitative end-point data? 3 What happens when stress is removed? Tina Toni ABC in systems biology 05/05/2011 9 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-30
SLIDE 30

Parameter estimation Application: Modeling bacterial stress response

Petri net model

H+

H+

H+

H+ Normal conditions Normal conditions ArcB PspA B C PspF Stress conditions ArcB PspA B C PspF

Stress membrane Damaged Intact membrane BcCcAc BCAF BCA

  • lg

F A BC 100 60 (40) 36 TF 6 6

Toni, Jovanovic, Huvet, Buck, Stumpf, BMC Systems Biology, 2011 (in press) Tina Toni ABC in systems biology 05/05/2011 10 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-31
SLIDE 31

Parameter estimation Application: Modeling bacterial stress response

Results: Possible qualitative behaviours

Stress induced: t = [0, 10), t = [30, 40). Stress removed: t = [10, 30)

dm

. . . . . . . . .

Toni, Jovanovic, Huvet, Buck, Stumpf, BMC Systems Biology, 2011 (in press) Tina Toni ABC in systems biology 05/05/2011 11 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-32
SLIDE 32

Parameter estimation Application: Modeling bacterial stress response

Results: Possible qualitative behaviours

Stress induced: t = [0, 10), t = [30, 40). Stress removed: t = [10, 30)

dm

. . . . . . . . .

Less severe stress → oscillations.

dm

. . . . . . . . .

Toni, Jovanovic, Huvet, Buck, Stumpf, BMC Systems Biology, 2011 (in press) Tina Toni ABC in systems biology 05/05/2011 11 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-33
SLIDE 33

Model selection

Bayesian model selection

candidate models

M1 M3 M2 M4 M5 M6

Marginal posterior distribution of a model

P(M|D) ∝ P(D|M)P(M) P(M|D) ∝

  • θ

P(D|M, θ)P(M)P(θ|M)dθ

Tina Toni ABC in systems biology 05/05/2011 12 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-34
SLIDE 34

Model selection

Bayesian model selection Fit to the data Number of parameters

Tina Toni ABC in systems biology 05/05/2011 13 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-35
SLIDE 35

Model selection ABC SMC for model selection

Model selection: P(M|D) =?

  • 1. Marginal likelihoods

For each model separately estimate P(D|M): P(D|M) =

  • θ

P(D|θ, M)P(θ|M)dθ Then P(M|D) = P(D|M)P(M)

  • M′ P(D|M′)P(M′).
  • 2. Joint space

Include model M as an extra parameter: (M, θ(1), . . . , θ(M)) ABC SMC → P(M, θ(1), . . . , θ(M)|D) Marginalize → P(M|D)

Tina Toni ABC in systems biology 05/05/2011 14 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-36
SLIDE 36

Model selection ABC SMC for model selection

Model selection on a joint space M1 M2 M3 M4

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-37
SLIDE 37

Model selection ABC SMC for model selection

Model selection on a joint space M1

M2

M3 M4

M∗

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-38
SLIDE 38

Model selection ABC SMC for model selection

Model selection on a joint space M1

M2

M3 M4

M∗ M∗∗ ∼ KM(M|M∗)

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-39
SLIDE 39

Model selection ABC SMC for model selection

Model selection on a joint space M1 M2

M3

M4

M∗ M∗∗ ∼ KM(M|M∗)

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-40
SLIDE 40

Model selection ABC SMC for model selection

Model selection on a joint space

(M3, θ1) (M3, θ2) (M3, θ3) (M3, θ4) (M3, θ6) (M3, θ7) (M3, θ8) (M3, θ9) (M3, θ5)

M∗ M∗∗ ∼ KM(M|M∗) θ∗

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-41
SLIDE 41

Model selection ABC SMC for model selection

Model selection on a joint space

(M3, θ1) (M3, θ2) (M3, θ3) (M3, θ4) (M3, θ6) (M3, θ7) (M3, θ8) (M3, θ9)

(M3, θ5)

M∗ M∗∗ ∼ KM(M|M∗) θ∗

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-42
SLIDE 42

Model selection ABC SMC for model selection

Model selection on a joint space

(M3, θ1) (M3, θ2) (M3, θ3) (M3, θ4) (M3, θ6) (M3, θ7) (M3, θ8) (M3, θ9)

(M3, θ5)

M∗ M∗∗ ∼ KM(M|M∗) θ∗ θ∗∗ ∼ KP(θ|θ∗)

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-43
SLIDE 43

Model selection ABC SMC for model selection

Model selection on a joint space

(M∗∗, θ∗∗)

M∗ M∗∗ ∼ KM(M|M∗) θ∗ θ∗∗ ∼ KP(θ|θ∗) accept / reject

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-44
SLIDE 44

Model selection ABC SMC for model selection

Model selection on a joint space

w(M∗∗, θ∗∗)

M∗ M∗∗ ∼ KM(M|M∗) θ∗ θ∗∗ ∼ KP(θ|θ∗) accept / reject calculate w

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 15 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-45
SLIDE 45

Model selection ABC SMC for model selection

Model selection on joint space: Weight calculation

wt(m∗∗, θ∗∗) = π(m∗∗, θ∗∗)

M

  • j=1

P(j)

t−1KMt(m∗∗|m(j) t−1)

  • model perturbation
  • k;mt−1=m∗∗

w(k)

t−1

  • l;mt−1=m∗∗ w(l)

t−1

KPt,m∗∗(θ∗∗|θ(k)

t−1)

  • parameter perturbation

(Toni and Stumpf, Bioinformatics, 2010) Tina Toni ABC in systems biology 05/05/2011 16 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-46
SLIDE 46

Model selection Application: Epo signaling pathway

Model selection: Epo signaling pathway

Epo hormone synthesised when oxygen levels low Regulates red blood cell production Deliver oxygen to tissue Epo = ”bloodbooster” (blood doping in endurance sports)

Tina Toni ABC in systems biology 05/05/2011 17 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-47
SLIDE 47

Model selection Application: Epo signaling pathway

Model selection: Epo signalling pathway

x1 x2 x3 x4

?

(adapted from Biocarta) Tina Toni ABC in systems biology 05/05/2011 18 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-48
SLIDE 48

Model selection Application: Epo signaling pathway

Model selection: Epo signalling pathway

1 2 3 Model 0.0 0.4 0.8 Population1 1 2 3 Model 0.0 0.4 0.8 Population2 1 2 3 Model 0.0 0.4 0.8 Population3 1 2 3 Model 0.0 0.4 0.8 Population4 1 2 3 Model 0.0 0.4 0.8 Population5 1 2 3 Model 0.0 0.4 0.8 Population6 1 2 3 Model 0.0 0.4 0.8 Population7 1 2 3 Model 0.0 0.4 0.8 Population8 1 2 3 Model 0.0 0.4 0.8 Population9 1 2 3 Model 0.0 0.4 0.8 Population10 1 2 3 Model 0.0 0.4 0.8 Population11 1 2 3 Model 0.0 0.4 0.8 Population12 1 2 3 Model 0.0 0.4 0.8 Population13 1 2 3 Model 0.0 0.4 0.8 Population14 1 2 3 Model 0.0 0.4 0.8 Population15

Tina Toni ABC in systems biology 05/05/2011 19 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-49
SLIDE 49

Model selection Application: Phosphorylation dynamics

Erk signaling pathway

Erk signaling pathway relevant to many cancers.

Tina Toni ABC in systems biology 05/05/2011 20 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-50
SLIDE 50

Model selection Application: Phosphorylation dynamics

Dual phosphorylation mechanisms

MAPKK MAPKK P MAPKK Disassociation P MAPKK P P MAPKK P P

Figure 1: Distributive phosphorylation.

MAPKK MAPKK P Bind and slide MAPKK P P MAPKK P P

Figure 2: Processive phosphorylation.

Tina Toni ABC in systems biology 05/05/2011 21 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-51
SLIDE 51

Model selection Application: Phosphorylation dynamics

Question

In vitro phosphorylation MAPK is distributive (Burack 1997, Ferrel 1997). In vitro de-phosphorylation MAPK is distributive (Zhao 2001). Is it the same in vivo? Cannot study phosphorylation and dephosphorylation separately.

Tina Toni ABC in systems biology 05/05/2011 22 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-52
SLIDE 52

Model selection Application: Phosphorylation dynamics

Question

In vitro phosphorylation MAPK is distributive (Burack 1997, Ferrel 1997). In vitro de-phosphorylation MAPK is distributive (Zhao 2001). Is it the same in vivo? Cannot study phosphorylation and dephosphorylation separately.

candidate models

D-D P-D P-P D-P

Tina Toni ABC in systems biology 05/05/2011 22 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-53
SLIDE 53

Model selection Application: Phosphorylation dynamics

Model selection: Phosphorylation dynamics

candidate models

D-D P-D P-P D-P

DD PP DP PD

Model

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 12

Mtot MKPtot Weights DD

5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 12

Mtot MKPtot Weights PP

5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 12

Mtot MKPtot Weights DP

5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 12

Mtot MKPtot Weights PD Tina Toni ABC in systems biology 05/05/2011 23 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-54
SLIDE 54

Model selection Application: Phosphorylation dynamics

Conclusion

ABC methods are highly applicable in systems biology

1

Prediction of dynamic behaviour

2

Testing hypothesis about biological systems

ABC SMC computationaly efficient

Tina Toni ABC in systems biology 05/05/2011 24 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011

slide-55
SLIDE 55

Acknowledgements

Acknowledgements

  • Prof. Michael Stumpf, IC
  • Dr. Goran Jovanovic, IC
  • Prof. Martin Buck, IC
  • Dr. Maxime Huvet, IC
  • Prof. Shinya Kuroda, University of Tokyo
  • Dr. Yuichi Ozaki, University of Tokyo

MRC Slovenian Academy of Sciences and Arts

  • Prof. Bruce Tidor, MIT

The Wellcome Trust

Tina Toni ABC in systems biology 05/05/2011 25 / 25

Nature Precedings : doi:10.1038/npre.2011.5964.1 : Posted 13 May 2011