Nouelling With Comsol: uiauients anu Pattein Foimation

Biiect questions anu suggestions to Benis (uzianis.menshykaubsse.ethz.ch)

Problem 1 Solve simple uiffusion equation (no ieactions!) on a 1B uomain.

In this problem you will learn very basics of Comsol usage: how to create a model
core and further populate it with model geometry and equations to be solved. You will
see how important to chose right mesh and eventually you will also see how species
diffuse on 1D domain.

Stait Comsol

First step is to create a core of the model, which include dimension of the model (1, 2
or 3D), type of equation to be solved as well as type of problem e.g. steady state or
time dependent.
MUDEL WIZARD
1. Click 1D button. To select 1 dimensional model.
2. Click Next {=>].
S. In the Add Pbysics tiee, select Cbemical Species Transport>Transport of
Diluted Species. This allow to solve PDE of reaction-diffusion type.
4. Click Add Selected {+].
S. Click Next {=>].
6. Choose Preset Studies for Pbysics>Time Dependent. This allow to solve time
dependent problem.
7. Click Finisb (F1 flag). Core of tbe moJel is creoteJ now!

Save youi file!

Second step is to enter parameters and their numerical values. It is a good practice in
programming and modeling to use constants with assigned numerical values instead
of entering directly numbers.

CLUBAL DEFINITIUNS
1. Right click Clobal Definitions.
2. Choose Parameters.
3. Type in Name: B0, Expression: 1F-10 [m^2,s]. Diffusion coefficient.
4. Type in Name: b0, Expression: 1F-4 [m]. Domain length.
5. Type in Name: c0, Expression: 1F-6 [mol,m^S]. Concentration of
morphogen at the boundary.
6. Type in Name: k0, Expression: 1E-8 [mol/(s*m^2)]. Morphogen flux at the
boundary.
7. Type in Name: moxt, Expression: 1uu|sj.

Third step is to specify domain on which equations should be solved.
CEUMETRY
1. Right click Ceometry.
2. Select Interval.
S. Select Interval 1.
4. Type in Right enupoint: b0.
S. Click Build.

Fourth step is to enter differential equations, boundary and initial conditions.
TRANSPURT UF DILUTED SPECIES
1. Click Convection and Diffusion.
2. Type in Biffusion coefficient: B0. In this case model has no reactions, only
diffusion.
S. Right click Transport Uf Diluted Species.
4. Click Concentration.
S. Click Concentration 1.
6. Select in Crapbics winuow point 1.
7. Click Add {+] in Concentration winuow.
8. Tick Species c.
9. Type in Cu,c: c0. Tbis specifies constont concentrotion ot tbe left bounJory of
tbe Jomoin. At tbe riqbt bounJory conJition remoins Jefoult {no flux).

Fifth step is to mesh computational domain. Selection of the right mesh is extremely
important if mesh is crude solution can easily might become inaccurate or wrong,
however selection of very fine mesh results in long computational times and increased
memory use.
Mesb
1. In Element size menu choose Extra fine.

Fventuolly moJel is reoJy!
Study
1. Click Step 1: Time Dependent.
2. Type in Times: ronqe{0,moxt,20,moxt).
S. Click Run {=].

If you typed everything right you should see following figure in Graphics section.

Save youi file!
Analyse giaph above anu answei following questions:
1. Bow concentiation is uistiibuteu at veiy shoit time.
2. Bow concentiation is uistiibuteu at long times.
S. What coulu you expect at veiy long times.

Nouify mouel to in the following way:
1. vaiy maximum time (change ulobal Definitions->Parameters->maxt).
2. Tiy uiffeient meshes (Mesb->Element Size). What happens if mesh is
veiy ciuue.
S. Change bounuaiy conuitions
a. Right click Transport of Diluted Species.
b. Click Flux.
c. Select in Crapbics winuow point 1.
u. Click Add {+] in Boundaries winuow.
e. Tick Species c.
f. Type in Cu,c: c0.
g. Change moxt to S s.

Problem 2 Nouel Noiphogen Binuing to Receptoi on a 1B uomain. (baseu on
Lanuei et all Bo Horpboqen 6roJients Arise by Biffusion? Bevelopmental Cell 2
(2uu2): 78S-796.)

ln tbis problem we continue to work on 1B Jomoin. Burinq tbis tutoriol you will
leorn bow to solve system of coupleJ PBFs of reoction-Jiffusionol type. You will olso
qet insiqbts into tbe morpboqen qroJients formotions, in porticulor, you will see
tbot JeqroJotion of receptor-morpboqen complex is strictly necessory for tbe
positionol informotion tronsJuctions. Iiqure below sbows scbeme of tbe consiJereJ
biocbemicol interoctions os well os tbeir formulotion in tbe form of PBFs.



MUDEL WIZARD
1. Click 1D button. To select 1 dimensional model.
2. Click Next {=>].
S. In the Add Pbysics tiee, select Cbemical Species Transport>Transport of
Diluted Species. This allow to solve PDE of reaction-diffusion type.
4. Click Add Selected {+].
S. Click Next {=>].
6. Choose Preset Studies for Pbysics>Time Dependent. This allow to solve time
dependent problem.
7. Click Finisb (F1 flag). Core of tbe moJel is creoteJ now!

Save youi file!

Befoie you go fuithei
Click Uptions->Preferences
In Model Builder tick Sbow More Uptions box

CLUBAL DEFINITIUNS
1. Right click Clobal Definitions.
2. Choose Parameters.
3. Type in Name: B0, Expression: 1F-11. Diffusion coefficient.
4. Type in Name: b0, Expression: 1F-4. Domain length.
S. Type in Name: moxt, Expression: 1000. Simulation time.
6. Type in Name: kon, Expression: 0.01. Rote constont of complex formotion.
7. Type in Name: koff, Expression: 1F-6. Rote constont of complex
Jissociotion.
8. Type in Name: c0, Expression: 1; Concentrotion ot tbe bounJory of tbe
Jomoin.
9. Type in Name: flux0, Expression: 1F-6; Ilux ot tbe bounJory of tbe
Jomoin.

CEUMETRY
1. Right click Ceometry.
2. Select Interval.
S. Select Interval 1.
4. Type in Right enupoint: b0.


TRANSPURT UF DILUTED SPECIES
1. Click Transport of Diluted Species.
2. Click Dependent.
a. Type in Number of Species: 2. ln tbis moJel we bove to voriobles -
concentrotion of morpboqen onJ concnentrotion of tbe bounJeJ
receptor.
b. Type in Concentrations: c.
c. Type in Concentrations: R.
S. Click Transport Mecbanisms.
a. 0ntick Convection. We consiJer Jiffusion only, no miqrotion of
cborqeJ species in electric fielJ or convection.
b. 0ntick Migration.
4. Click Consistent Stabilization
a. 0ntil All boxes.
S. Click Convection and Diffusion.
a. Type in c: B0.
b. Type in R: 0. lt is known tbot receptors usuolly Jiffuse 100 - 10000
times slower tbon morpboqens, bere we ossume tbot receptor is not
Jiffusinq ot oll.
6. Click Initial Values 1.
a. Type in c: 0. lnitiolly we ossume no morpboqen or receptor present.
b. Type in R: 0.
7. Right click Transport of Diluted Species. Now we oJJ cbemicol reoctions
Jescribinq liqonJ-receptor interoctions.
8. Select Reactions.
9. Click Reactions 1.
a. Select in Crapbics winuow line 1. Eere we specify Jomoin wbere
reoctions ore boppeninq.
b. Click Add {+] in Domains winuow.
c. Type in : -kon*c*{1-R)-koff*R; Type in reoctions for tbe liqonJ
onJ receptor.
u. Type in : kon*c*{1-R)-koff*R;
1u. Right click Transport of Diluted Species. No enter bounJory conJitions.
11. Select Flux.
12. Click Flux 1.
a. Select in Crapbics winuow point 1.
b. Click Add {+] in Concentration winuow
c. Tick Species c box.
u. Type in : flux0.
e. Type in Cb,c: cu.

Mesb
2. In Element size menu choose Extra fine.

Study
1. Click Step 1: Time Dependent.
2. Type in Times: ronqe{0,moxt,20,moxt).
S. Click Run {=].

Results ln results section specify type of plot you woulJ like to see os well os cboose
vorioble you woulJ like to plot.
1. Click Line grapb 1 in 1D Plot Croup.
2. Right click 1D Plot Croup 2.
a. In Y-axis type in Expression: R.
b. Click Plot.
Now you sboulJ see plot similor to tbot below, wbicb sbows tbot witb time more
onJ more receptors become populoteJ. So ofter sufficiently lonq time positionol
informotion con not be reoJ out.

lt is known tbot receptor os well os otber proteins JeqroJe in cells, so we oJJ
JeqroJotion term next to see bow it influence receptor Jistribution.
To Bo:
1. Incluue uegiauation of ieceptoi
a. Select Reactions 1 fiom Transport of Diluted Species.
b. Change Rr fielu to: kR*c*{1-R)-koff*R-kJeq*R.
c. Auu kJeq=0.01 to Parameters at Clobal Definitions.
You shoulu see figuie similai to that below:

lf JeqroJotion of bounJeJ receptor present stip qroJient is observeJ
even ot lonq times, so cells con reoJ out informotion onJ
Jifferenciote.
2. vaiy kJeq paiametei value, which impact uoes it have on concentiation
piofile.
S. vaiy oiuei of the uegiauation ieaction:
a. Change Rr fielu to: kR*c*{1-R)-koff*R-kJeq*R^n
b. n can be u, 1, 2
c. Impoitance of self enhanceu uegiauation is auuiesseu in the
papei: Eluai A et oll Self-FnbonceJ liqonJ BeqroJotion0nJerlies
Robustness of Horpboqen 6roJients Bevelopmental Cell
S(2uuS):6SS-646.
4. vaiy othei paiameteis values, which impact uoes it have on calculateu
concentiation piofiles of moiphogen anu ieceptoi.

Problem 3. Cell vaiiability anu giauient foimation on a 2B uomain (baseu on
Bollenbach et all Piecision of the Bpp giauient Bevelopment 1SS (2uu8), 11S7-
1146).

ln tbis problem we stuJy qroJient formotion on o 2B Jomoin. ln previous exercises
we stuJieJ formotion of qroJients on 1B, bere we extenJ our stuJies in to seconJ
Jimension. We olso consiJer cell voriobility, wbicb result in Jiffusion coefficient,
JeqroJotion onJ proJuction rotes beinq o function of spotiol coorJinotes. Eere
insteoJ of Tronsport of BiluteJ Species interfoce we use qenerol Hotbemoticol
interfoce, wbicb proviJes more flexibility.

MUDEL WIZARD
1. Click 2D button. To select 2 dimensional model.
2. Click Next {=>].
S. In the Add Pbysics tiee, select Matbematics>PDE Interfaces>Ceneral Form
PDE. This allows to solve PDE of reaction-diffusion type.
4. Click Add Selected {+].
S. Click Next {=>].
6. Choose Preset Studies for Pbysics>Time Dependent. This allows to solve
time dependent problem.
7. Click Finisb (F1 flag). Core of tbe moJel is creoteJ now!

Save youi file!

Befoie you go fuithei
Click Uptions->Preferences
In Model Builder tick Sbow More Uptions box

CLUBAL DEFINITIUNS
Beie all paiameteis aie uimensionless.
1. Right click Clobal Definitions.
2. Choose Parameters.
a. Type in Name: B, Expression: 1. Biffusion coefficient.
b. Type in Name: b0, Expression: 1. Domain length.
c. Type in Name: l0, Expression: 1. Domain width.
u. Type in Name: moxt, Expression: 10. Simulation time.
e. Type in Name: k, Expression: 0.S. BeqroJotion rote constont.
f. Type in Name: flux, Expression: 1. Horpboqen flux ot tbe
bounJory.
g. Type in Name: no, Expression: 0. Noise omplituJe.
S. Right click Clobal Definitions.
4. Choose Functions>Random.
a. Name: noiseB
b. Number of arguments: 2
c. Standard Deviation: no.
S. Choose Functions>Random.
a. Name: noisek
b. Number of arguments: 2
c. Standard Deviation: no.
6. Choose Functions>Random.
a. Name: noisef
b. Number of arguments: 1
c. Standard Deviation: no.

CEUMETRY
1. Right click Ceometry.
2. Select Rectangle.
S. Widtb: l0.
4. Heigbt: b0.

PDE
1. Click PDE.
2. Click Dependent.
a. Type in Number of dependent variables: 1.
b. Type in Concentrations: u.
S. Click PDE.
4. Click Ceneral Form PDE
a. Conservative Flux x: -ux*B*{1-noiseB{x,y)).
b. Conservative Flux y: -uy*B*{1-noiseB{x,y)).
c. Source: -u*k*{1-noisek{x, y)).
u. Damping of Mass Coefficient: 1.
e. Mass Coefficient: 1
S. Click Initial Values
a. Nake suie u anu its ueiivative aie set to zeio.
6. Right click PDE anu select FluxJSource.
a. Select left siue of the iectangulai anu click plus.
b. Boundary FluxJSource g: flux*{1-noisef{y)).

Mesb
1. In Element size menu choose Extra fine. lf occurote results requireJ mesb
size sboulJ be siqnificontly smoller tbon ony qeometricol feoture or feoture
in tbe solution.

Study
1. Click Step 1: Time Dependent.
2. Type in Times: ronqe{0,moxt,10,moxT).
S. Click Run {=].

ColculoteJ Jistribution of morpboqen is similor to colculoteJ in Problem 2, now
switcb on noise by settinq no=0.4 onJ Run moJel oqoin.

Results Now we will moke cross section 1B plots
1. Right click Date Sets anu select Cut Line 2D.
2. Click newly cieateu Cut Line 2D submenu.
S. Befine cioss-section lines
a. Point 1: 0, 0.1
b. Point 2: 1, 0.1
c. Tick box Additional parallel lines
u. Distance: iange(u,u.1,u.S).
4. Right click Results anu select 1D plot group
S. Right click newly cieateu 1D plot group anu select Line grapb
6. In 1D plot group in Data set uiop uown menu select Cut Line 2D.
7. Plot cioss-section giaph.

You con see tbot bowever porometers of moJel siqnificontly Jeviote from tbe meon
colculoteJ morpboqen qroJients ore olmost inJepenJent of tbe noise level. Tbis
sbows tbot even in biosystems witb biqb level of noise onJ voriobility in porometers
morpboqen qroJients proviJes relioble informotion for spotiol position reoJ out.

To Bo:
Inciease paiametei no which contiols noise amplituue, see to which extenu
moiphogen uistiibution is affecteu by the noise. Check influence of cell
vaiiability only in a souice anu only in a sink.
Fuithei ieauing:

1. ueneial
a. Piof Ibei lectuie notes
http:www.bsse.ethz.chcobieuucationNath_Nou_basic
b. Nathematical Biology by }ames B Nuiiay
2. Noiphogen giauients:
a. Lanuei A. et oll Bo Horpboqen 6roJients Arise by Biffusion?
Bevelopmental Cell 2 (2uu2): 78S-796.
b. Eluai A. et oll Self-FnbonceJ liqonJ BeqroJotion0nJerlies
Robustness of Horpboqen 6roJients Bevelopmental Cell
S(2uuS):6SS-646.
c. uiegoi T. et oll Piobing the Limits to Positional Infoimation Cell
1Su(2uu7):1SS-164.
u. Kicheva A. Kinetics of Noiphogen uiauient Foimation Science
S1S(2uu7):S21-S2S.
e. Bollenbach et all Piecision of the Bpp giauient Bevelopment 1SS
(2uu8), 11S7-1146
f. Piof. Philip Naini web-site (http:people.maths.ox.ac.ukmaini)
S. Comsol
a. Biowse thiough available mouels at
http:www.comsol.comshowioom
b. Reau manuals)


Questions anu suggestions aie welcomeu; uiiect them to
uzianis.menshykaubsse.ethz.ch