theoretical models in immunology - cbs · theoretical models in immunology ... predator/preys...

30
CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS BiC BioCentrum-DTU Technical University of Denmark Theoretical models in immunology Nicolas Rapin Immunological bioinformatics Center for Biological Sequence analysis.

Upload: phamxuyen

Post on 01-Sep-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Theoretical models in immunology

Nicolas RapinImmunological bioinformatics

Center for Biological Sequence analysis.

Page 2: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Why models?

Models make simplifications of the reality, mainly an unknown reality.

A model can describe different parts of the reality, its function or its appearance.

Models have different levels of abstraction.

Page 3: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Page 4: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Two kinds of mathematical models

Descriptive models, where the aim is to fit biological data to achieve prediction through a statistical model calibration. ‣development of experimental design techniques.

Analytic models, based on systems of ordinary differential equations (ODEs), take into consideration the mechanisms involved in the studied system.‣a famous example is the Lotka-Volterra model for predator/preys interactions.

Page 5: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

A very simple model

The variable P increases at rate k per unit of time.

The ODE describes the change in P.k is a free parameter.

example: bank account.P

k

Page 6: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

constant growth rate : dP/dt = k

0 5 10 15 20 25

5

10

15

dy/dt = k

solution: y(t) = k*t

dP/dt= k - s = k’

k

k’

Page 7: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Exponential decay

The variable P decreases at a rate proportional to the current value of P (times a factor k).

The solution is well known as:

examples are decay of radioactive particles.

P k

Page 8: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Exponential decay: dy/dt = -k*y

0 5 10 15 20 25

5

10

15

dy/dt = -d*y

solution: y(t) = Y0*exp(-k*t)

Half life is calculated as follow

Here, all the curves have the same k, but different starting points

Page 9: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Half life and life span

Half life is the time it takes for a given population to be halved.

Life span is the inverse, 1/half life.

so if a cell in a culture live on average for 2 days, the half life of the cell population is:

1/2 = 0,5 days.

Page 10: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Exponential growth

The variable P change at a rate proportional to the current value of P (times a factor n).

The solution is well known as:

P

n

Page 11: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Fitting bacterial growth:

‣The experiment:

‣The assumptions derived from experiments:‣ The population doubles every hour. ‣ On average, each Bacteria enter mitosis every hour.

‣Can one predict the population size at a given time point?

Time(h) 0 1 2 3 4

# bacteria 10 20 40 80 160

Page 12: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Fitting bacterial growth:

0 2.5 5 7.5

400

800

1200

1600

2000

2400

0 2.5 5 7.5

400

800

1200

1600

2000

2400

n= 0.69 = ln(2)/1growth rate is 69% per hour.

Page 13: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

A first realistic model

We change make the assumption that the more money you have in the bank, the more you spend, whatever you get from your earning.

Will you ever become rich?

k

dP

Page 14: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

constant growth - exp. decay

0 5 10 15 20 25

5

10

15

dy/dt = k - d*y

solution: y(t) = k/d*(1-exp(-dt))+Y0*exp(-dt)

dy/dt = k-dY_= 0 => Y_ = k/d

this is the steady state of the system

Page 15: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Mass action law

Page 16: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Interacting populations

The predator/prey system:Rabbit population grows exponentially, and is

a source of food for pumas.Puma population increases by eating rabbits,

and decrease following exp. decay.

R

n

d

a

P

Page 17: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Interacting populations

0 2500 5000 7500 1!104

1.25!104

40

80

120

160

200

240

Rabbits numbers

Pumasnumbers

Stable limit cycle

Page 18: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Interacting populations

Target cells TInfected cells I

Pumas and rabbits can be replaced by immunologicaly relevant entities, such as target cells and infected cells in the case of viral infection.

T

v

b

I dIdT

Page 19: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Logistic growth

Exponential growth doesn’t always reflect reality completely.

n

dP

Two solutions:‣Density dependent death rate and fixed growth

rate.‣Density dependent growth rate and fixed death

rate.

Page 20: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Density-dependent rates

Maximal rate

indi

vidu

al b

irth

/dea

th ra

te

Population size

Death rate

Birth rate

Minimal rate

indi

vidu

al b

irth

/dea

th ra

tePopulation size

Death rate

Birth rate

equilibrium point

carrying capacity

Page 21: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Density-dependent rates

Density independent death rate:rate is always

Density-dependent death rate:rate is a (linear) function of the population

Page 22: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Logistic growth

The model becomes:

0 2.5 5 7.5 10 12.5 15

250

500

750

1000

1250

1500

K = 1250

Exponential growthdP/dt = rN

Logistic growthdP/dt = rN (1- P/K )

Page 23: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Page 24: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Cellular automata

‣The environment is a matrix like a chess

board. (Lattice)

‣Each cell has a “state”

‣On, Off, Alive, Infected, Dead...

‣Update of the state of a cell defined by rules

according to neighborhood.

‣Not continuous, works step by step.

Page 25: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

The game of life

‣Each cell has 8 "neighbors", which are the cells adjacent to it.

‣Each cell can be either alive or dead.‣A dead cell with exactly 3 live neighbors becomes

alive (or is born). ‣A live cell with 2 or 3 live neighbors stays alive;

otherwise it dies (from loneliness or overcrowding).

Page 26: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Small example

Who in the class says the state changes?

Page 27: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

The game of life

Page 28: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

The future, agent based models

NaiveTh0

ActivatedTh0

Th1

helperT cell

Th2

IL12

IL4

TNFαTGFβ

NaiveB

ActivatedB

Plasmocyte

INFγ

MF

IgM

IgG1

IgG2

(class switchof B cell)

(class switchof B cell)

becomes(1)

(1)

(1)

(1)

(2)

(2)

(2)

(2)

secretes

affects (+/-)

contact

LEGEND

(+)

(+)

(+)(+)

(+) (-)(-)

(-) (-)

Input

Input

more states, movements, diffusion even complete organs and much more!!

Page 29: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Page 30: Theoretical models in immunology - CBS · Theoretical models in immunology ... predator/preys interactions. ... ‣The environment is a matrix like a chess board. (Lattice)

CEN

TER FOR BIO

LOG

ICA

L SEQU

ENC

E AN

ALY

SIS

BiC BioCentrum-DTUTechnical University of Denmark

Parameter estimation

Parameters value need to:

‣Be estimated from experiments.‣Guessed.

Mathematical model allow to explore the parameter space or test some assumptions.

It si best to have few parameters & processes than many!