Anda di halaman 1dari 24

Approximation of a cell cycle model

in Budding Yeast
• Introduction
• Cell Cycle Model of Budding Yeast
• Constant Input Response (CIR) of the Full Model
• Proposed Method
p g Jacobian-based Local Refinement ((JLR))
• Preprocessing:
• Initialization of the Genetic Algorithm (GA)
• Results and Discussions
• Comparison between 2 Scenarios for Substitution of Excluded States
• Results of Substitution Excluded States with Mean Values
• GA with Parameter Estimation
• Results of the GA with Parameter Estimation
• Conclusion and Future Works

Presented by Chaiyut Thanukaew 1


Introduction

• Molecular mechanisms in biological systems are complex and have dynamic


interactions.
• To express behaviors of a system, often a set of nonlinear ODEs is occupied.
• Since normally huge models, mathematical model reduction is used to gain insight
into the models and reduce computational cost.
• There are 36 ODEs, 25 algebraic equations and 139 parameters in the cell cycle
model of budding yeast.
• The reduced model must retain essential features of the full model, traditionally,
the trajectories of certain state variables. Moreover, it must comply with viability
criteria.
• For the cell cycle model of budding yeast, responses of the mass regarding
external input is decided to be the preserved characteristic.

 Objective: to minimize the number of states with preserving certain characteristics


of the model by using genetic algorithm.

2
Cell Cycle Model of Budding Yeast (1)

Spindle 4p
phases of eukaryotic*
y cell cycle;
y ; G1-,, S-,, G2-
Nuclear Formation and M-phase.
Migration
• G1 or Gap1 phase: the cell grows.
G2 • S or DNA synthesis: the cell makes
Chromosome
copies of its chromosomes and get ready
M Segregation,
DNA Nuclear to divide.
S
Replication Division
• G2 or Gap2 phase: the cell check the
duplicated chromosomes.
Bud
G1 • M or Mitosis phase: the cell separates the
Emergence
Cytokinesis copies chromosomes into 2 full sets and
Spindle (Cell Division) the cell divides into 2 new cells
Pole, Body
Duplication Start Growth y
How to know which mechanisms of the cell cycle
take place at what time?
 CDK-Cyclin** is a protein active in regulating
Figure 1: Cell Cycle of Budding Yeast
y
the cell cycle.

* Eukaryote is a cell with visible nucleus e.g. yeast and mammal. 3


** Cyclin Dependent Kinase and Cyclin
Cell Cycle Model of Budding Yeast (2)

Viability Criteria
• Correct sequences, i.e.
Budding  DNA synthesis
 Mitosis  Chromosome
segregation  Cell
division.
• [mass] <= 10 all time.

Figure 2: Consensus Model of Budding Yeast


4
Constant Input Response (CIR) of Full Model (1)
CIR definition: “For each added constant input, the CIR is the amplitude and time of the
mass state such that it becomes stable while, also, the whole model must comply the
viability criteria”.

Constant Input added to the state 5 (Sic1);


d[Sic1]
 (k ' s , c1  k" s , c1.[Swi5])  (Vd , b 2  kdi, b 2).[C 2]  (Vd , b5  kdi, b5).[C5]  kpp, c1.[Cdc14].[Sic1P]
dt
 (kas, b 2.[[Clb2]  kas, b5.[[Clb5]  Vkp ) [Sic1]  K added
k , c1).[

Added constant input

Why adding at [Sic1]?


 As Sic1 is a regular cyclin subunit called stoichiometric CDK inhibitor, directly adding
a constant signal, physically, would mean putting some substances to boost up some
activities of the cell cycle, e.g. going faster into the G1-phase.

How to notice the response of the budding yeast model after adding up the constant
signal to the Sic1 state?
 The easiest answer is to measure how the mass physically reacts to this
perturbation.
5
Constant Input Response (CIR) of Full Model (2)
Response of the mass to different constant inputs

Mass Stable
Kadded Time Amplitude

0.035 ~ 520 ~3
0.15 ~ 190 ~5
0.245 ~ 460 ~8
0.30 Inviable

6
Figure 3: The mass state when different constant inputs are applied
Proposed Method (1)
The full model is written as; We rewrite the full model as;
 
x  f ( x, p ) x  f ( x, s , p )

Where; x: the vector of states


p: the vector of parameters, and
s: a vector where si is “1” when the i-th state is included in the model
((“0”
0 if excluded).
excluded)
Objective

The reduced model x  f ( x, s * , p) is defined such that its solution x* minimizes
the cost function  , i.e.

 reduction ( x  )  min  reduction ( x)


sS

Where;  : the cost function, and


S : th
the sett off allll vectors
t off length
l th Nf (Number
(N b off states
t t off the
th ffullll model)
d l)
whose entries are “0” or “1”. 7
Proposed Method (2)
Cost Function
The cost function  is defined as;

 1 X is not oscillatoryy or inviable.



 
1.  1.
 CIR X is oscillatory and viable.

 Not count reduced models with negative cost as feasible solutions.

Where;

The relative model size  is,

Nr the number of states in the reduced model



Nf the number of states in the full model

8
Proposed Method (3)
Where (continued);

The CIR error is computed as the least square distance between one of the
full and one of the reduced model as following;
N per

 ( xF (i )  xR (i )) 2  ( y F (i )  y R (i )) 2
 CIR  i 1
N per


i 1
xF2 (i )  y F2 (i )

• xF and yF represent the positions of the CIR of the ffullll model when
hen
different constant inputs are applied,
• Nper is the maximum number of different constant inputs for which the
full model is still oscillatory and viable
viable, and
• xR and yR represent the positions of the CIR for the reduced model
regarding different constant inputs added.

9
Proposed Method (4)

Generation “0”
0 Generation “n”

Do linear ranking of all


Randomly generate an individuals
individual
Copy 2 best individuals
Jacobian-based local from the previous
refinement (JLR) generation (Elitism)
Replacement
p
Random selection 2
Calculate cost of each (till reaching
parents acc. to their ranks
individual setting “n” or
satisfied cost
Crossover and mutation
value or no
Total Ni individuals
improvement
Jacobian-based local
after some time)
refinement (JLR)

Calculate cost of each


Figure 4: Flowchart of the GA offspring
Applying to Reduce the Model
Total Ni individuals 10
Preprocessing : JLR (1)
Jacobian matrix and JLR
 The matrix of all first-order partial derivatives of a vector-valued function returning
relationship between equations in a system regarding that vector-valued function.
 JLR uses information from Jacobian matrix for local refinement.

Considering the following equation;


~
J  J  diag
di (J(J )
Where, J: the Jacobian matrix, and
diag(J): all diagonal elements of the Jacobian matrix
matrix.
~
If there is any row in J such that all elements are zeros  input-only state, and
~
If there is anyy column in J such that all elements are zeros  output-only
p y state.

 Can leave out the input- and output-only states because of redundancy, e.g. the input-
only state can be substituted by a constant and the output-only state can be indirectly
obtained
bt i d b by some other
th states.
t t
11
Preprocessing : JLR (2)
Index table of the original cell cycle model

Input-only
states

~
Table 1: Index Table J  J  diag ( J ) of the Cell Cycle Model Output-only 12
states
Initialization of the GA (1)

Problem: Very hard to obtain individuals being oscillatory and viable in the gen. “0”.
Reasons would be as following;
• Generating individuals by rounding a random number of uniform distribution
causing that half of the number of bits is “0” in average (18 from 36 bits). There is
the problem, e.g. assume that oscillatory individuals must comprise of at least 26
states, thus it is hard to obtain this number.
• The generated individuals lack of some crucial states, e.g. mass and some states of
cyclins. Thus, leaving out only one of these states causes the reduced models non-
oscillatory or inviable.

Solution  introduce preconditions to generates the individuals in the initial generation


of the GA.

13
Initialization of the GA (3)
Example: Feasible individual generated in gen. “0” with preconditions

bit1 1  State 1: set to “1” (mass)


bit 2 R
   State 3 and 4: randomly generated
bit 3 R
 
bit 4 1  State 4: set to “1” (viability concerned)
bit 5 1  State 5: set to “1” (constant input added)
 
  
 
 
bit 26 L  State 26, 27: likely set to “1s” (only missing
bit 27 L
  one state,
t t non-oscillatory)
ill t )
 
 
 
bit 33 1
 
bit 34 1
bit 35 1 State 33, 34, 35 and 36: set to “1s”
  (viability concerned)
bit 36  1 

Figure 5: Example of Generating a Feasible Individual in Gen. “0” 14


Comparison between Zero VS Mean-valued Sub.
2 scenarios of substitution an excluded state;
• Substitution with zero: the excluded states is completely cut away, and
• Substitution with mean value: the excluded state still has influence, but not
dynamic behavior
behavior.

Case #1 Case #2 Case #3

Representation Binary bit string Binary bit string Binary bit string

Population Size 80 100 80

Fitness Assignment Linear ranking Linear ranking Linear ranking

Selection Fitness-based Fitness-based Fitness-based

Crossover Uniform Uniform 2-point

Mutation Bit mutation prob.=0.1 Bit mutation prob.=0.1 Bit mutation prob.=0.1

Replacement Scheme Whole population Whole population Whole population


replacement with elitist replacement with elitist replacement with elitist
no =2
no.=2 no =2
no.=2 no =2
no.=2

15
Table 2: Parameter Setting of 2 Substitution Scenarios
Results : Comparison between Zero VS Mean-valued Sub.
The best result regarding
g g Table 3 comes from the mean-valued substitution

2. Reduced model with mean-valued


substitution

Case #1 Case #2 Case #3

Total State No. 23 24 25

Total Cost 0.7244 0.7521 0.7754

Table 3: Results of Mean-valued Substitution

 Model reduction
red ction with
ith mean
mean-valued
al ed substitution
s bstit tion is accredited for ffurther
rther
simulations since it returns the better cost value (the lower value).

 The total state number can be subtracted 3 because the states 34, 35, 36 are
output-only states added to determine the viability (see slide 12).

 Not obvious seen difference between each case of the same substitution
scenario even population size and crossover strategy are varied
varied.
16
Results : Mean-valued Substitution (1)
Table 4: Results of Mean-valued
Substitution from Several Simulations

Purpose:  to prove that the results


do not come from local minima
minima.

• The best reduced models “S1” and


“S2” contains 23 states.
• Although, different number of
states for each result, all results are
somehow in the same pattern (blue
highlight).
highlight)
• If fixing highlight areas, the best
result can be obtained from varying
of 10 bits in non
non-highlighted
highlighted areas
(search space is reduced).
• Further test “S1” and “S2”
comparing their responses to one of
the full model .
17
Results : Mean-valued Substitution (2)
The cell cycle of the full model VS the cell cycle of the reduced model

Fi
Figure 6
6: C
Cellll C
Cycle
l off th
the F
Fullll M
Model
d l VS C
Cellll C
Cycle
l off “S1” iin T
Table
bl 4
18
 Cell cycle of “S2” is very similar to “S1” except that Clb2T is higher (max. nearly 1.5).
Results : Mean-valued Substitution (4)
Reduced model “S1” as the Jacobian index table

: Excluded
states (non-
dinamic)
: Substituted
by constant
: Non-zero
Jacobian

19
Table 5: Reduced Model “S1” Substituting in the Jacobian Index Table
Results : Mean-valued Substitution (4)
Reduced model “S1” as the Jacobian index table

Table 6: Jacobian Index Table of the Reduced Model “S1”

Note: State 24 (Tem1) is input-only state but cannot be left out since it 20
involves resetting parameters.
GA with Parameter Estimation (1)

Idea: number of feasible solutions might be increased if the thresholds of determination


the viability criteria are changed.
Implementation method: to put 4 more bits representing real value numbers used as the
thresholds of the viability determination
determination.

Representation 36 Binary bit strings and 4 positive real values

Population size 80

Fitness assignment Linear ranking

Selection Fitness-based

Crossover Uniform crossover with prob. = 0.5 (crossover in the parameter part is
substituted by the median values of the parents)

Mutation Bit mutation with prob. = 0.1 and normal distribution added with the
previous value in the parameter part

Replacement Whole population replacement with elitist no. = 2


Scheme

Table 7: Parameter Setting of the GA with Parameter Estimation 21


Results: GA with Parameter Estimation

Results of the GA with Parameter Estimation

Reduced Model with Discussions


Parameter Estimation The results are not better than the results with no
Round#3 Round#3 parameter estimation. The reasons would be as
Ind.1 Ind.2 following;
Kez ((0.3)) 0.4572 0.4242 pp p
• Inappropriate estimation of “Thres. 1”
Kez2 (0.2) 0.1720 0.2485 because it is used as the threshold for 3 states
[ORI], [BUD] and [SPN],
Thres. 0.1 0.1050 0.0806
• Newlyy estimated parameters
p cannot lead
Thres. 1 0.6481 0.9425 feasible individuals into optimal region in the
search space, and
Total State 24 25
No. • New pparameters are ggenerated by y the normal
Total Cost 0.7765 0.7954 distribution with a small variance.

Table
T bl 88:The
Th 2 B
Bestt R
Resultlt off GA with
ith P
Parameter
t
Estimation (from 3 times of simulation) 22
Conclusion and Further Works
• The best result contains 23 states. 3 more states can be ignored because of the
output-only states.
• Appropriate GA’s parameters might increase efficiency of the simulation, e.g. lessen
the computational
p time,, explore
p the search space
p intelligently.
g y

• The GA hybridizes with the Jacobian-based local refinement can reduce


unnecessary computation, e.g. the GA with JLR in this work.

• Although GA can be applied to resolve a given problem without prior knowledge of


such that problem, initialization of the GA is applied to this problem because of
physical reality and model constraints.
• Further works would be to study more about the GA with parameter estimation and to
combine other strategies of model reduction to the GA, e.g. to lump 2 continuous states
together.

1 2 3 1 2 3

Thank You 23
Appendix A : CIR of the Full Model

Figure AA.1:
1: CIR of the full model such Figure A
A.2:
2: CIR of the full model such
that Kadded is increased 0.01 each that Kadded is the median values of
time (from 0 to 0.28) each group

24

Anda mungkin juga menyukai