where computer science meets biology

57
Where Computer Science meets Biology Systems Biology Professor Adriana Compagnoni Computer Science Department With: Joe Glavy, Svetlana Sukhishvili, Tommy Wh Amanda DiGuilio (CCBBME) Yifei Bao, Vishakha Sharma, Justin Sousa, Peter Zafonte (CS)

Upload: rich

Post on 13-Jan-2016

36 views

Category:

Documents


3 download

DESCRIPTION

Systems Biology. Where Computer Science meets Biology. Professor Adriana Compagnoni Computer Science Department. With: Joe Glavy, Svetlana Sukhishvili, Tommy White Amanda DiGuilio (CCBBME) Yifei Bao, Vishakha Sharma, Justin Sousa, and Peter Zafonte (CS). Overview. The Virtual Lab - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Where Computer Science meets Biology

Where Computer Science meets Biology

Systems Biology

Professor Adriana CompagnoniComputer Science Department

With: Joe Glavy, Svetlana Sukhishvili, Tommy WhiteAmanda DiGuilio (CCBBME)

Yifei Bao, Vishakha Sharma, Justin Sousa, and Peter Zafonte (CS)

Page 2: Where Computer Science meets Biology

OverviewOverview

• The Virtual Lab

Modeling HER2

• Modeling Bio-films

• Experimenting with 5 different modeling techniques

Modeling the G-protein Cycle

• Incorporating Space

Page 3: Where Computer Science meets Biology

Process model for HER2 Signaling Pathways

Page 4: Where Computer Science meets Biology

HER2HER2

Page 5: Where Computer Science meets Biology

HER2 – Signaling Pathways HER signaling is initiated when two receptor

molecules join together in a process called dimerization, which occurs in response to the presence of a growth factor molecule (ligand).

Dimerization of different members of ErbB family, lead to activation of different kinase mediated intracellular signaling cascades, like PI3 K/AKT pathway MAPK pathway JAK/STAT pathway

Page 6: Where Computer Science meets Biology

Process Algebra ModelProcess Algebra Model

Page 7: Where Computer Science meets Biology

Overview of different speciesOverview of different species

Page 8: Where Computer Science meets Biology

Differential HER2 concentrationsDifferential HER2 concentrations

a. 100 HER2 are initialized a. 30 HER2 are initialized

Page 9: Where Computer Science meets Biology

Inhibition of Raf activation(activation rate is changed from 27.0 to 1.0E-4 )Inhibition of Raf activation(activation rate is changed from 27.0 to 1.0E-4 )

Pathways Crosstalk

Page 10: Where Computer Science meets Biology

Computational Modeling of Bio-films

for Drug Delivery

Computational Modeling of Bio-films

for Drug Delivery

Page 11: Where Computer Science meets Biology

Hydrogen-bonded multilayer destruction

Changing pH

3.2 μm

3.2 μm

fast release of capsule cargo

Page 12: Where Computer Science meets Biology

Computational Modeling For The G-protein Cycle

Computational Modeling For The G-protein Cycle

12

Using 5 different modeling techniques:

• Differential equations• Stochastic Pi-calculus• Stochastic Petri-Nets• Kappa• Bio-Pepa

Page 13: Where Computer Science meets Biology

G-Protein Couple Receptors(GPCRs)G-Protein Couple

Receptors(GPCRs)

13

• G-protein couple receptors (GPCRs) sense molecules outside the cell and activate inside signal transduction pathways.

• An estimated 50% of the current pharmaceuticals target GPCRs.

Page 14: Where Computer Science meets Biology

Activation cycle of G-proteins by G-protein-coupled receptors

Activation cycle of G-proteins by G-protein-coupled receptors

14

1

2

3

45

Page 15: Where Computer Science meets Biology

Chemical Reactions and RatesChemical Reactions and Rates

15

Page 16: Where Computer Science meets Biology

The law of mass actionThe law of mass action

16

The law of mass action is a mathematical model that prescribes the evolution of a chemical system in terms of changes of concentrations of the chemical species over time.

In its simplest form, it says that a reaction

X+Yk Z has a rate k[X][Y]: the rate is proportional to the concentration of one species ([X]) times the concentration of the other species ([Y]) by the base rate k.

Page 17: Where Computer Science meets Biology

Ordinary Differential Equations (ODEs)

Ordinary Differential Equations (ODEs)

17

Page 18: Where Computer Science meets Biology

Process AlgebrasAn alternative ot ODEsProcess Algebras

An alternative ot ODEs

Formal languages originally designed to model complex reactive computer systems.

Because of the similarities between reactive computer systems and biological systems, process algebra have recently been used to model biological systems.

Formal languages originally designed to model complex reactive computer systems.

Because of the similarities between reactive computer systems and biological systems, process algebra have recently been used to model biological systems.

18

Page 19: Where Computer Science meets Biology

Process AlgebrasProcess Algebras

Typically two halves of a communication: Sending and Receiving.

!ch(msg) : to send message msg on channel ch.

?ch(msg) : to receive a message msg on channel ch.

Typically two halves of a communication: Sending and Receiving.

!ch(msg) : to send message msg on channel ch.

?ch(msg) : to receive a message msg on channel ch.

19

Page 20: Where Computer Science meets Biology

Process AlgebrasProcess Algebras

A and B bind

A and B dissociate

A and B bind

A and B dissociate

20

A = new msg !ch (msg); Ab(msg) B = ?ch(msg); Bb (msg)

Ab(msg) = !msg(); A Bb(msg) = ?msg(); B

Page 21: Where Computer Science meets Biology

Stochastic Pi-CalculusStochastic Pi-Calculus

The stochastic pi-calculus is a process algebra where stochastic rates are imposed on processes, allowing a more accurate description of biological processes.

SPiM (stochastic pi machine) is an implementation of the stochastic pi-calculus that can be used to run in-silico simulations that display the change over time in the populations of the different species of the system being modeled.

The stochastic pi-calculus is a process algebra where stochastic rates are imposed on processes, allowing a more accurate description of biological processes.

SPiM (stochastic pi machine) is an implementation of the stochastic pi-calculus that can be used to run in-silico simulations that display the change over time in the populations of the different species of the system being modeled.

Page 22: Where Computer Science meets Biology

Step 1 to Step 2 Step 1 to Step 2

22

1

2

Gd

Gbg

G!bindb

?bindb

Gd represents alphaGbg represents the beta-gamma complex

Page 23: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

23

Page 24: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

24

Page 25: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

25

Page 26: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

26

Page 27: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

27

Page 28: Where Computer Science meets Biology

Process modeling for G-protein CycleProcess modeling for G-protein Cycle

28

Page 29: Where Computer Science meets Biology

SPiMSPiMdirective plot RL(); abrD(); aT()

Page 30: Where Computer Science meets Biology

Petri Nets ModelingPetri Nets Modeling

The basic Petri Net is a directed bipartite graph with two kinds of nodes which are either places or transitions and directed arcs which connect nodes. In modeling biological processes, place nodes represent molecular species and transition nodes represent reactions. We use Cell Illustrator to develop our model of the G-protein cycle (www.cellillustrator.com).

The basic Petri Net is a directed bipartite graph with two kinds of nodes which are either places or transitions and directed arcs which connect nodes. In modeling biological processes, place nodes represent molecular species and transition nodes represent reactions. We use Cell Illustrator to develop our model of the G-protein cycle (www.cellillustrator.com).

30

Page 31: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 32: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 33: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 34: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 35: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 36: Where Computer Science meets Biology

NJPLS2010@Stevens

Petri Nets ModelingPetri Nets Modeling

Page 37: Where Computer Science meets Biology

Kappa Language ModelingKappa Language Modeling Kappa is a formal language for defining agents

(typically meant to represent proteins) as sets of sites that constitute abstract resources for interaction. It is used to express rules of interactions between proteins characterized by discrete modification and binding states. The Kappa language modeling platform is Cellucidate.

Kappa is a formal language for defining agents (typically meant to represent proteins) as sets of sites that constitute abstract resources for interaction. It is used to express rules of interactions between proteins characterized by discrete modification and binding states. The Kappa language modeling platform is Cellucidate.

Page 38: Where Computer Science meets Biology

Kappa Language ModelingKappa Language Modeling

R and L are agent names, r is the binding site of R, l is the binding site of L, and 3.32e-09 is the reaction rate. In R(r!1) and L(l!1), 1 is the index of the link that binds R and L at their binding sites.

R and L are agent names, r is the binding site of R, l is the binding site of L, and 3.32e-09 is the reaction rate. In R(r!1) and L(l!1), 1 is the index of the link that binds R and L at their binding sites.

R + L -> RLR(r), L(l)->R(r!1), L(l!1) @ 3.32e-09

Page 39: Where Computer Science meets Biology

Kappa Language ModelingKappa Language Modeling

Page 40: Where Computer Science meets Biology

Bio-PEPA ModelingBio-PEPA Modeling Bio-PEPA is a process algebra for the modeling

and the analysis of biochemical networks. It is a modification of PEPA to deal with some features of biological models, such as stoichiometry and the use of generic kinetic laws.

Bio-PEPA is a process algebra for the modeling and the analysis of biochemical networks. It is a modification of PEPA to deal with some features of biological models, such as stoichiometry and the use of generic kinetic laws.

Page 41: Where Computer Science meets Biology

Bio-PEPA ModelingBio-PEPA Modeling

Page 42: Where Computer Science meets Biology

Bio-PEPA ModelingBio-PEPA Modeling

Page 43: Where Computer Science meets Biology

Result (ODEs and Pi-calculus)Result (ODEs and Pi-calculus)

Page 44: Where Computer Science meets Biology

Result (Petri Nets and Kappa)Result (Petri Nets and Kappa)

Page 45: Where Computer Science meets Biology

Result (Bio-PEPA)Result (Bio-PEPA)

Page 46: Where Computer Science meets Biology

High Level Notations (motivation)

High Level Notations (motivation)

Both ODEs and Petri Nets correspond closely to chemical reactions, and for the average biologist, they are relatively easy to understand.

Cellucidate provides a friendly user interface for Kappa that abstracts away from its syntax.

SPiM still needs encoding in the stochastic Pi-calculus.

Both ODEs and Petri Nets correspond closely to chemical reactions, and for the average biologist, they are relatively easy to understand.

Cellucidate provides a friendly user interface for Kappa that abstracts away from its syntax.

SPiM still needs encoding in the stochastic Pi-calculus.

Page 47: Where Computer Science meets Biology

High Level Notation (motivation)High Level Notation (motivation)

In order to hide Pi-calculus communication primitives and enable modeling using only terminology directly obtained from biological processes, we develop a high level notation that can be systematically translated into SPiM programs.

In order to hide Pi-calculus communication primitives and enable modeling using only terminology directly obtained from biological processes, we develop a high level notation that can be systematically translated into SPiM programs.

Page 48: Where Computer Science meets Biology

G-protein CycleG-protein Cycle

48

1

2

3

45

bind bind

activate and dissociate

dissociate

hydrolyze

Page 49: Where Computer Science meets Biology

Step 1 to Step 2 Step 1 to Step 2

49

1

2

bind(Gd, Gbg, G, 1.0)

Page 50: Where Computer Science meets Biology

High Level NotationHigh Level Notation

bind(Gd, Gbg, G, 1.0); bind(R, L, RL, 3.32e-6); activateAnddissociate(G, RL, Ga, Gbg, 1.0e-5); dissociate(RL, R, L, 0.01); hydrolyze(Ga, Gd, 0.11); degrade(R, 4e-4); degrade(RL, 4e-3)

bind(Gd, Gbg, G, 1.0); bind(R, L, RL, 3.32e-6); activateAnddissociate(G, RL, Ga, Gbg, 1.0e-5); dissociate(RL, R, L, 0.01); hydrolyze(Ga, Gd, 0.11); degrade(R, 4e-4); degrade(RL, 4e-3)

Page 51: Where Computer Science meets Biology

High Level NotationHigh Level Notation

Page 52: Where Computer Science meets Biology

High Level NotationHigh Level Notation

Page 53: Where Computer Science meets Biology

We are currently modifying the Ocaml implementation of SPiM to Add a notion of Space (Affine Geometry + Process Algebra)

Incorporating SpaceIncorporating Space

Page 54: Where Computer Science meets Biology

ConclusionsConclusions

The models we build using stochastic modeling approaches can represent the G-protein cycle quite convincingly, which shows that stochastic modeling approaches could be efficient instruments to assist in biomedical research.

In fact, because of the randomness of dynamic biological systems, stochastic modeling approaches can make the description of biological process much simpler and more accurate.

The models we build using stochastic modeling approaches can represent the G-protein cycle quite convincingly, which shows that stochastic modeling approaches could be efficient instruments to assist in biomedical research.

In fact, because of the randomness of dynamic biological systems, stochastic modeling approaches can make the description of biological process much simpler and more accurate.

Page 55: Where Computer Science meets Biology

ConclusionConclusion

The high level notation that we designed is a domain specific notation that we developed for the G-protein cycle.

The high level notation that we designed is a domain specific notation that we developed for the G-protein cycle.

Page 56: Where Computer Science meets Biology

Thank you!Questions?

Thank you!Questions?

Page 57: Where Computer Science meets Biology

OverviewOverview

G-protein-coupled Receptor G-protein Cycle Ordinary Differential Equations (ODEs)

Modeling Stochastic Pi-Calculus Modeling Petri Nets Modeling Kappa Language Modeling High Level Notation

G-protein-coupled Receptor G-protein Cycle Ordinary Differential Equations (ODEs)

Modeling Stochastic Pi-Calculus Modeling Petri Nets Modeling Kappa Language Modeling High Level Notation

57