1

Part IBiological Basis of Systems Biology

Systems Biology: Advances in Molecular Biology and Medicine, First Edition. Edited by Robert A. Meyers.© 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

3

1Systems Biology

Melanie Boerries1, Roland Eils2, and Hauke Busch1

1Freiburg Institute for Advanced Studies – LifeNet, School of Life Sciences,Albertstraße 19, 79104 Freiburg, Germany2German Cancer Research Institute, Im Neuenheimer Feld 280, 69120Heidelberg, Germany

1 Introduction 5

2 What Is Systems Understanding? 6

3 Why Are Biological Systems Different? 83.1 Biological Complexity 83.2 Global Properties of Biological Systems 103.2.1 Robustness of Biological Systems 103.2.2 System Adaptation and Control 113.2.3 Modules and Protocols 12

4 Systems Biology Modeling 134.1 Network Biology 164.2 Dynamic Network Models 174.3 Reaction–Diffusion Models 184.4 Holism versus Reductionism: The Global Dynamics of Networks 204.5 Modeling Resources and Standards 20

5 Future Prospects of Systems Biology 225.1 Synthetic Biology 225.2 Conclusions: Where Are We? 23

References 25

Systems Biology: Advances in Molecular Biology and Medicine, First Edition. Edited by Robert A. Meyers.© 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

4 Systems Biology

Keywords

Systems biology

A new field of biology that studies the functional structure and dynamics of intercellular

and intracellular networks with the help signal- and systems-oriented methods.

Synthetic biology

Studies life as networks of biological objects such as DNA proteins RNA and metabolites.

Network biology

Studies the static organization of life as networks made up of biological entities such as

DNA proteins RNA or metabolites.

System

A set of interacting parts functioning as a whole and distinguishable from its

surroundings by identifiable boundaries.

Systems theory

This denotes the cross-disciplinary investigation of the abstract organization of systems

irrespective of their substance type or spatiotemporal scale of existence. The goal is the

study of emerging properties that arise from the interconnectedness of the individual

parts making up the system.

Robustness

The robustness of biological systems denotes the maintenance of specific system

functionalities in the presence of fluctuations or change in environmental parameters.

Control

Control is defined as the response action taken by a system to counteract parameter

changes to maintain system functions at a certain predefined level.

Modularity

A design concept of complex systems to integrate simpler self-contained functional

building-blocks into the framework of one larger system.

Model

The concept of representing causal relationships from real systems in the language of

mathematics.

Systems Biology 5

Systems Biology is a new field of biology, which places the theoretical foundationsof systems analysis of living matter into the context of modern high-throughputquantitative experimental data, mathematics, and in silico simulations. The aim is toanalyze the organization and to gain engineering-control of metabolic and geneticpathways. The ultimate goal is to gain an ‘‘holistic’’ view of the complex workingsof life. The need for a system level understanding of biology is reviewed in thischapter, and comments are provided on the current scientific progress in this field.The current and future directions of experimental design strategies and theoreticalapproaches are also highlighted.

1Introduction

Systems Biology is a recently establishedfield in life sciences that aims at promotinga global systems understanding of livingmatter through the integration of variousscientific domains (see Refs [1–3] forspecial journal sections and Refs [4–7] fortextbooks on the topic).

The considerate attention that SystemsBiology receives is due to the fact that itcurrently causes a paradigmatic shift inmany areas of biological research. Modernmolecular biology has been mostly a de-scriptive science, devoting insight to small,isolated compartments of a system as awhole – for example, by investigating theinfluence of individual proteins within thebehavior of a whole cell. Thus, the study ofthe interconnected nature of cellular pro-cesses has long been avoided in favor ofa reductionist approach. On the one handthis is due to the sheer amount of newchallenges that come about when tacklingcomplex systems, whereas on the otherhand it has been the common leitmotif inother natural sciences, such as physics,to shed light mostly on well-definedand controllable systems. Those systemsare then either small and isolated, orlarge and homogeneous, so that theycan be tackled by applying the laws ofstatistics. Novel challenges, however, lie in

the description of dynamical, mesoscopic,open, spatiotemporally extended, nonlin-ear systems, operating far away from ther-modynamic equilibrium, which are themost important type to understand, asthese are the systems that support life.

Although reductionist approaches havebeen successful in elucidating key pro-cesses and key factors of many funda-mentally important biological processes,contemporary science is now realizingthe importance of wholeness by studyingproblems of organization. Emergent phe-nomena arise from the interaction of var-ious units or modules, which are neitherresolvable nor understandable through thestudy of local events or the respectiveparts in isolation. Hence, traditional re-ductionist models and methods of celland molecular biology are not very wellsuited, and can be incomplete, misleading,or even completely wrong.

Historically, Jan Christiaan Smuts wasamong the first to formulate a theory ofthe whole that was hoped to fill the gapbetween science and philosophy. In hisbook Holism and Evolution, which waspublished in 1926 [8], Smuts argued thatNature consisted of discrete objects, or‘‘wholes,’’ that are not entirely resolvableinto their respective parts. The wholesand parts mutually depend on each otherin their functionality, thus forming oneorganic, unified web of relations, which

6 Systems Biology

comprises matter, life, and mind, andwhich cannot be accounted for by areductionistic analysis. Smuts saw his ideaconfirmed in evolution, regarding Holismas the active driving force towards moreperfect wholes or species.

The theoretical foundations of systemsengineering were laid some 60 years ago,when the concept of systems theory inbiology was proposed during the 1940sby the biologist Ludwig von Bertalanffy[9]. The proposal was further developedduring the 1950s by Ross Ashby [10], as acounter-movement against reductionismin science. In the sense of holism, vonBertalanffy emphasized the need for astudy of the informational organizationwithin real, open systems. The assembly ofsuch inter-related elements then comprisea unified whole, which in turn can shownew emergent properties.

In 1948, the mathematician NorbertWiener established the field of cybernetics[11] as the science of communication andcontrol of systems in regard to their en-vironment. Cybernetics is closely relatedto systems theory, using the same con-cepts of information, control, or feedback.However, whereas the former focuses onsystems function for providing regularand reproducible behavior, the latter dealsmore with system structure. Even so, bothterms are often used in conjunction, forboth structure and function cannot beunderstood as separate entities.

Today, biology embarks on systemsthinking in two different ways. One wayis to regard Systems Biology as a new waytoward integrating information from dif-ferent organizational levels, starting fromDNA to proteins via signaling pathwaysto functional modules, into the contextof a holistic organizational view [12]. Theprimary goal of the second view on Sys-tems Biology is to establish a conceptual

framework and working methodologiesfor the augmentation of knowledge on bio-logical phenomena by combining systemstheory and molecular biology: ‘‘SystemsBiology is not a collection of facts but away of thinking’’ [13]. This view has al-ready been shared during the late 1960sby Mesarovic, who predicted that SystemsBiology would be an established field ofscience as soon as ‘‘. . . biologists startasking the right questions’’ [14]. Put dif-ferently, biologists need not recast factsalready known from molecular biology ina different language, but they need toask questions based on system-theoreticconcepts [15].

Both approaches share the extensiveneed for high-quality, quantitative bio-logical data obtained through extensiveexperimental measurements, and this isthe reason why the use of systems theoryin biology has gained momentum duringrecent years. New techniques can pro-vide the necessary amount of quantitativedata for the establishment of appropriateholistic models of cellular processes. Even-tually, those new experimental techniqueswill lay the foundation for the integrationof mathematics, engineering, physics, andcomputer science into biology, to permitan understanding of the range of complexbiological regulatory systems at multiplehierarchical and spatiotemporal levels ofcellular organization [16].

2What Is Systems Understanding?

The word ‘‘system’’ derives from the Greekσνστημα, and is composed of the prefixsyn, which means ‘‘together,’’ and the rootof histanai, meaning ‘‘cause to stand.’’ Asystem is defined as ‘‘. . . the assembly orset of inter-related elements comprising

Systems Biology 7

a unified whole that is distinct from itsenvironment,’’ and can be hierarchicallyorganized and made up of other subsys-tems or modules, which allows the con-struction of a complex entity from simplerunits. For example, organelles such as mi-tochondria constitute distinct subsystemswithin the organization of a cell. The sub-division of natural entities into systems isan abstract construct. Systems per se donot really exist in reality; rather, they aredefined as a set of elements interactingover time and space.

Systems theory denotes the transdisci-plinary investigation of the abstract or-ganization of phenomena, independentof their substance, type, or spatiotempo-ral scale of existence [17]. The goal ofsystems theory is to study emerging prop-erties arising from the interconnectednessand complexity of relationships betweenparts. Such theory argues that howevercomplex or diverse a system is, there arealways different types of organizationalstructures present, which can be repre-sented as a network of information flow.Because these concepts and principles re-main the same across different scientificdisciplines such as biology, physics, or en-gineering, systems theory can provide abasis for their unification. The systemsview distinguishes itself from the moretraditional analytic approach by emphasiz-ing the concepts of system–environmentboundaries, signal input–output relation-ships, signal and information processing,system states, control, and hierarchies.Albeit systems theory is valid for all sys-tem types, it usually focuses on complex,adaptive, self-regulating systems which aretermed ‘‘cybernetic.’’

Elegant, simple, and globally valid mod-els are rare in biology as compared toother fields of science. Few examples existwhere a function can be attributed to the

workings of a single small molecule or fewproteins, as in the case of hemoglobin forthe transport of gases in the bloodstream,or bacterial chemotaxis [18].

In general, many genes and proteinsare involved in cellular responses to ex-ternal stimuli. In general, biology followsa reductionist approach by investigatingsmall, isolated parts of a cell, tissue, or or-ganism; typically, biology tries to deducebiological phenomena from molecular be-havior, which often results in a simplistic‘‘one gene for one function’’ approach.However, since genetic analysis has shownthat the genotypes of different species aremostly identical, it would appear that it isthe signal processing stage(s) on the wayfrom the genome to the phenotype – inother words, an ever more elaborate reg-ulation of gene expression [19] – whichcarries the subtle particularities in the re-spective genetic codes.

As a consequence, biological phenom-ena should be explained within thevocabulary of system theory, such as am-plification, control, adaptation, sensitivity,autoregulation, and error correction, tak-ing an holistic view of the system underconsideration [20]. In short, Systems Biol-ogy is required to uncover the laws of thewhole that cannot be inferred by delvingdeeper into the details.

The systematic investigation of biolog-ical matter comprises the understandingand control of system structure and dy-namics in the sense of systems theory andcybernetics, respectively [1].

System structure denotes the identifica-tion of the static connection topology andregulatory relationships within the net-work of genes, proteins, and other smallmolecules that constitute the signal trans-duction and metabolic pathways, as wellas the physical structure of organisms orcells. Experimental techniques to elucidate

8 Systems Biology

the cells’ global system structure includefor example DNA microarrays [21], deepsequencing [22], and protein–protein in-teraction screens via yeast-two-hybrid [23]or split-ubiquitin approaches [24]. Systemdynamics refers to the qualitative andquantitative evolution of the above networkover time. Dynamics include the tempo-ral variation of molecule concentrations,as well as the structure of the network it-self. Examples of experimental techniquesto study cellular system dynamics in-clude fluorescent imaging techniques tomonitor molecular dynamics and inter-actions on the level of the individualcell [25]. Moreover, mass spectrometry ortranscriptomics can be used to investigatethe collective behavior of the proteome orgene expression in cells over time.

3Why Are Biological Systems Different?

While biological systems must still bebased on the laws of physics, chemistry,and thermodynamics [26], biology also in-corporates the notion of organismal func-tion. This notion represents the need forsurvival and reproduction, as well as thepossibility to evolve in and adapt to chang-ing environments, and it is this inherentpurpose that distinguishes biology signifi-cantly from all other natural sciences [27].Moreover, there is no distinct separation ofinformation storage and regulatory units.Genes, for example, can regulate their ownexpression by gene splicing [28].

Classical physics views the emergenceof every effect to be determined by a causeresiding in the past. Biological systems,on the contrary, are teleonomic – that is,they are oriented towards a state in thefuture [17]. It was the insight of cyberneticsthat purposeful activity can be described

through the use of circular mechanisms,where the effect equals the cause. Thesimplest example of a circular logic is afeedback loop, in which the output of thesystem is fed into its input again (seeSect. 3.2).

Having these prerequisites in mind, theconcepts of systems theory and cybernet-ics should be utilized, if there is a desireto establish successful formal mathemati-cal models in biology. The introduction ofcircular causalities has far-reaching conse-quences on the general design and globalproperties of biological systems, such asrobustness, complexity, and control, allof which are discussed in detail in thischapter.

3.1Biological Complexity

In order to comprehend the challengesin Systems Biology, it is important firstto understand the origin of the complex-ity encountered in Nature. For once, thecomplexity within a system does not nec-essarily come about as a consequence ofthe number of its component. In physics,the macroscopic state or the dynamicsof objects are often well described by afew parameters or simple mathematicalequations. For example, although the tem-perature of an object comes about dueto the thermal agitation of its individualatoms, the average energy of the atomscan be ascribed to a single, macroscopi-cally measurable, quantity.

A complex system is one in which thelaws that describe its behavior are qual-itatively different from those that governits units, such that new features emergewhen moving from one temporal, spatial,or organizational scale to another. The sci-ence of complexity is about revealing theprinciples that govern the ways in which

Systems Biology 9

these new properties appear [29]. Thus, itis rather the organization of the systeminto irreducible, heterogeneous parts thatare highly structured and hierarchically or-ganized on various spatiotemporal scales,which makes a system become complexor, in other words, ‘‘complicated’’ [30].

The notion of complexity is not welldefined amongst the various science disci-plines. Information theory usually char-acterizes complexity as the amount ofinformation needed to optimally predictthe behavior (or state) of the system basedon entropy measures [31]. Adami uses se-quence complexity of biological genomesto define the amount of information storedabout the environment [32]. Another viableapproach toward defining the complexityof biological systems may be the identifica-tion of the topological structure at a higherlevel of large-scale organization in termsof hierarchically organized networks [33].

Complex systems can show the emer-gence of ordered macroscopic behavior,termed self-organization. While biologicalsystems are ideal candidates to demon-strate self-organization, the applicability ofits principles to elucidate biological phe-nomena is, to date, rather limited. Thisis due mainly to the fact that biologicalsystems require a re-definition of com-plexity that is quite different from thatin physics. Biological systems are hetero-geneous, modular, highly structured onmultiple irreducible spatiotemporal scales,and also self-dissimilar, with each en-tity usually having several functional andregulatory properties. Nevertheless, theindividual components ‘‘. . . interact selec-tively and nonlinearly to produce coherentrather than complex behavior’’ [34].

One explanation for the emergence ofself-organization is the so-called ‘‘slaving-principle.’’ Under certain conditions theglobal, macroscopic behavior of a system

is governed by a few, slowly evolvingstate variables, which ‘‘slave’’ all otherdynamics. In this way, the relevant de-grees of freedom are largely reduced,allowing the system to find its own struc-ture – that is, to self-organize [35]. One useof the slaving principle is the analyticaldescription of lasers, explaining the spon-taneous, synchronous light emission bythe atoms in the lasing medium. Anotherframework for the explanation of emer-gent phenomena in multibody systemsis the notion of self-organized criticality(SOC) [36], which was acclaimed to ex-plain the frequent occurrence of long-taildistributions in many natural phenomena.SOC denotes the ability of open and dis-sipative systems to display critical – thatis, scale-invariant – behavior even in theabsence of external pressures. However,whilst this theory did not live up to itspromises of being universally applicable,it is applied in earthquake, forest fire, oravalanche models.

Carlson and Doyle introduced the the-ory of highly optimized tolerance (HOT)[37], which accounts for the intrinsic de-sign of biological and engineered systems.Such theory reflects the behavior of suchhigh-throughput, high-density systems,which are faced with limited resources.These systems show a high tolerance (i.e.,robustness) against environmental param-eter fluctuations, with the robustness be-ing achieved at the cost of a high degree ofcomplexity through the addition of controlunits to the system. Resource constraintsthen call for an optimized trade-off be-tween fault tolerance towards frequentlyexperienced perturbations and fragility torare, yet possible, events. Thus, there existsa ‘‘conservation law of robustness’’ [38].

The complex design of the living cell isoften compared to the make-up of comput-ers or today’s commercial aircraft. These

10 Systems Biology

man-made machines have a bewilderingcomplexity, and no man alone under-stands all the parts and their interplayin complete detail. Many of the stronglyinteracting and irreducible complex func-tions are there to automate, control, orback-up operations.

These definitions elucidate the maindifferences between systems describableby SOC or HOT; the former theory relatesto scale-invariant and self-similar features,while the latter deals with self-dissimilarstructures and sensitivities (dis-)appearingon each scale of observation, such that thedegrees of freedom in these systems areirreducibly many.

3.2Global Properties of Biological Systems

Typically, systems thinking seeks univer-sal properties for biological systems, link-ing the emergence of complexity to generaldesign features. Such features can be usedas a guideline for abstract mathemati-cal modeling by examining the principlesthat are commonly shared between diversespecies. These properties, as discussed be-low, show how biology and engineeringconverge on a systems level view.

3.2.1 Robustness of Biological SystemsThe fault tolerance or robustness of bio-logical systems denotes the maintenanceof specific system functionalities, even inthe presence of fluctuations or a changein environmental parameters [39, 40]. Inbiology, this refers to the concept of home-ostasis and the stability of developmentalcontrol. Robustness is achieved throughthe incorporation of regulatory controlloops into a system, thus shielding orbuffering the desired system functionalityfrom environmental influences. Robust-ness is a relative system property: in order

to maintain a certain equilibrium, otherproperties must evolve and adapt. Thereis, therefore, a need to define which cellu-lar functions change, and which resist asa reaction towards a disturbance.

Most of the ‘‘complicatedness’’ encoun-tered in biological systems is a directconsequence of the implementation ofcontrol schemes, rather than the core func-tion itself. In particular, if robustness isto be achieved for a wide range of distur-bances, the control must have an equallyincreasing variety, as stated in the ‘‘Lawof Requisite Variety’’ [10]. Consequently,complexity serves simplicity in the sensethat the complicated control schemes arehidden underneath the simple, yet reliable,output [41].

Robustness in biological systems is usu-ally achieved by various strategies, includ-ing redundant or functionally overlappingregulatory pathways [42], feedback loopsto regulate signal responses [43], or check-points within the cell [44, 45]. All of thesecontrol elements add to the total numberof components regulating cellular signal-ing elements, while keeping the numberof phenotypic expression levels low [20].Cellular checkpoints play important roles,such as in mitosis [45, 46], in the cell cycle[47], and during the early stages of embryodevelopment [48]. Bacterial chemotaxis isan intensively studied example of a robustbehavior due to multiple feedback con-trol [49, 50]. Examples of genetic bufferingto achieve robustness are shown by thefact that only about 20% of genes in bud-ding yeast are essential for viability [51].To a certain extent, genetic buffering cansafeguard against environmental changesor genetic defects. The simplest bacteria(e.g., Mycoplasma pneumonia) may haveonly a couple of hundred genes yet, despitetheir unexpectedly complex transcriptomeorganization (which includes antisense

Systems Biology 11

transcripts, alternative transcripts, andmultiple regulators per gene [52]) theysurvive only within a narrow band of en-vironmental parameters. In contrast, thebacterium Escherichia coli possesses about3000 genes, and is able to survive undera variety of environmental conditions [30]by activating additional control schemesunder environmental stress.

A further consequence of robustness isthat dynamical behavior in biological sys-tems is coupled less to the parametersthemselves, and more to the overall sys-tem structure. Previously, Von Dassow [53]showed that robustness was the simplify-ing criterion for determining the correcttopology of the segment polarity network,producing a highly robust patterning overa large range of parameter variation.

Robustness in complex systems comesat the price of some fragility due totiny, yet rare, events. Indeed, a small re-arrangement of some cellular signal path-ways can often lead to a spectacularfailure – that is, impaired cell death. Forexample, although cancer may be causedthrough an accumulation of mutationsin the genetic code over the human lifespan [54], the occurrence of cancer dur-ing the reproductive years of a humanis rather rare. It appears that Naturecan cope well with this trade-off for thebenefit of robust and reliable ‘‘normal’’functioning.

3.2.2 System Adaptation and ControlAdaptation is the ability of a system toaccommodate for varying external inputstimuli or disturbances in order to gainand maintain the correct and optimizedoutput; this is usually achieved with thehelp of feedback control. A well-studiedexample of adaptation using integral feed-back control is chemotaxis, in which bac-teria adapt their movement not to the

absolute level of pheromones but ratherto the chemical gradient, only [18, 50, 55].Another example is the activation of heatshock proteins in E. coli under temperaturestress [56], which function as chaperonesfor correct protein folding and the preven-tion of unwanted protein aggregation.

Control is a recurrent design encoun-tered in natural systems, which effec-tively increases robustness, with the goalof keeping certain values within prede-fined physiological limits. It is possibleto distinguish between feed-forward andfeed-back system controls; the former is anopen-loop sequence of predefined actionstriggered by a certain stimulus, workingdependably only within strict ranges ofinput stimuli, yet simple in design [57],while the latter employs a closed-loopdesign, which feeds part of the outputsignal back into the system input. De-pending on the sign of the feed-back,positive feed-back (or autocatalysis) willamplify the output signal, enhancing re-liable sensitivity toward cellular decisionderived from noisy input signals [58],whereas negative feedback usually sta-bilizes the output around some desiredvalue by opposing any changes causedby disturbances. For example, a combi-nation of feed-back control schemes canbe used to stabilize the receptor presen-tation of a cell to increase its sensitivitytoward a broad range of external lig-and concentrations; this was demonstratedin the case of erythropoietin receptorsignaling [59].

Although both types of control aresuccessfully employed in intracellularsignaling [60], both also bear fragilities.Autocatalysis can lead to self-sustained,unrestricted signal amplification, as ob-served in uncontrolled tumor growth [61],while ultra-stable homeostasis regulationcan cause large, possibly disturbing,

12 Systems Biology

transient signals in response to externalfluctuations [38].

Most often, biological systems usecombinations of open- and closed-loopdesigns, basing their control action onthe absolute, accumulative or differentialvalue of the input stimulus, in this waybalancing both sensitivity and stability.

3.2.3 Modules and ProtocolsModules are subsystems characterized asmostly self-contained entities with fixedinterfaces for external communication.They are evolutionarily detached, possess-ing their own identity, and have manyinternal – but only few external – links forinformation and matter exchange [27].

Modularity seems to be an importantconcept in biology, probably stemmingfrom the evolutionary pressure for optimalflexibility and the low chance of damagespread. Spatially distinct entities, such asorganelles in the cell or metabolic units,appear to be a recurrent scheme foundthroughout various organisms. The generegulatory network in cells seems to beorganized in modules [62]. Interestingly,modular design is a commonly found prin-ciple in modern engineering, as it enablesthe independent development and testingof units before their integration into a com-mon system. The benefits comprise sav-ings in developmental and maintenancecosts, as well as the prospect of gracefuldegradation rather than catastrophic fail-ure, as errors are usually restricted to asingle module.

If the modular design of biological sys-tems is a governing principle, this opensup new possibilities for the simplificationof in vivo and in silico experiments. Mod-ules would provide fixed levels of detailand size, which are easily abstracted oncetheir functions are known. Moreover, withthe help of these core modules it would

be possible to build ever-more complexmodels, without the need for segmenta-tion of every level of detail [62].

Modules communicate by using proto-cols; these are fixed, commonly agreed-onrules that standardize communicationwith their respective surroundings. In ad-dition, protocols ensure error correction,cellular coordination, and also evolvabil-ity through the possibility of adding newfunctions to a particular module. Protocolshave been shown to be an efficient meansfor ensuring the hierarchical organizationof complex systems through their integra-tion of different layers, thus reducing thecosts of information transmission [63]. Forexample, negative feed-back is a powerfulmodule for the establishment of home-ostasis. Similarly, gene regulation, mem-brane potentials, or signal transductionpathways can all be regarded as protocolswhich are utilized by different biologicalmodules, such as the DNA, ion channelsor kinases, and phosphatases [38]. Elabo-rate feed-back control protocols have beenrobustly established for the spatiotempo-ral development of various species [60,64]. An extensive genome-wide analysis ofprokaryotic cell-cycle progression has re-vealed a hierarchical control structure withthree to four master regulatory proteinsacting in a coordinated fashion [65].

Interestingly, the use of protocols on acellular level has striking similarities withthe way that modern computers communi-cate via the internet. The internet employsmany protocols for persistent communi-cation that can be ordered into a generalhierarchy, starting from the data link viathe network and transport to the appli-cation layer. Popular examples of eachhierarchy are the reverse address resolu-tion protocol (RARP) for the IP (internetprotocol) look-up and search of the com-munication partner, the IP for the routing

Systems Biology 13

for the appropriate subnet of the data,the TCP (transmission control protocol)for the secure, error-free data exchangeand, finally, the application layer such asthe HTTP (hypertext transfer protocol) orFTP (file transfer protocol) for retrievinga web page or downloading files. The factthat these protocols have been used fordecades, irrespective of the rapid and on-going evolution of computer hardware andsoftware, highlights their importance.

4Systems Biology Modeling

The word ‘‘model’’ derives from the Latinmodus – that is, ‘‘manner’’ or ‘‘measure,’’and refers to the concept of represent-ing causal relationships from real systemsin the language of mathematics. Thismapping often involves simplifications ofthe original systems, with the hope ofgaining predictive power on experimentalresults and explaining functional designprinciples. Life is an emergent property

stemming from the interaction amongmolecules, and cannot be reduced to theindividual properties of the molecules.Modeling tries to infer and predict therelations between molecules in terms ofcausation – that is, it tries to establishexplanatory relationships of the spatiotem-poral changes of matter [66].

Modeling and simulation have becomeindispensable tools for gaining insightsinto natural systems. In biology, they helpto bridge the gap between theory andexperiment. Often, biologists are facedwith the dilemma that experiments donot provide sufficient data for theoreticalinterpretation, while at the same time,clues for new experiments are missing asa consequence of lacking hypotheses [67].

Experimental results require correctmathematical interpretation, and modelhypotheses require experimental proofs[68]. The process of knowledge generationis iterative in nature, and consists of twofeedback loops (see Fig. 1). Here, the ex-perimental part employs high-throughputtechniques to obtain quantitative data

Model

Methods Observations

KnowledgeGeneration

Hypothesis

ExperimentalDesign

Predictions

Match ?

No

Yes

Fig. 1 Schematic to the experiment–modeling loop forknowledge generation in Systems Biology.

14 Systems Biology

relating to the system dynamics; thedata acquired are then compared withpredictions obtained from the mathemati-cal model. In the case of failure – that is, inthe case of diverging results from modeland experiment – this process must be re-peated by adjusting the old hypothesis andthus reconciling the model with the newresults from the experiment.

As an example of this cycle of modelbuilding, experiment, and model refine-ment, Ideker et al. [69] have recently devel-oped an integrated approach to construct,test, confirm and refine the simulationof the yeast galactose utilization network,through the combined numerical simula-tion and analysis of networks with system-atic experimental perturbation and globalmeasurements.

The major guideline for optimal mod-eling should be the concept of Occam’srazor, which states that models should bevoid of any redundant information. Yet to-day, this paradigm needs also to be viewedfrom the opposite aspect – models mustnot be oversimplified, so as to miss theessential clues of functionality of real lifesystems. In particular, it is important tonote that modeling in biology must es-tablish different concepts from those inphysics. The paradigm of nonlinear sci-ence is that even simple systems can leadto a dynamically rich behavior. Despite thebeauty of the simplicity of this idea, how-ever, it must be realized that living systemsregulate their dynamics differently – thatis, through a complex make-up of intercon-nected regulatory functions, a fact whichis often neglected or ignored [70]. Mod-els in biology are usually heuristic; theyarise embedded in the process of biolog-ical experiments, and are coupled tightlyto them. Information is, for example, de-duced from a perturbation analysis of theexperimental system, and thus contains

assumptions regarding the causality andthe passage of time [70].

In general, experimental data mustbe comprehensive with respect to fouraspects [57]:

• Factor: the need to capture the behaviorof all important target factors, such asgenes and proteins that play decisiveroles in the experimental system underconsideration.

• Item: this refers to the simultaneousmeasurement of the necessary sets ofvariables that are required for reliablehypothesis building, such as transcrip-tion level, molecule concentration, orspatial location.

• Time and space: this refers to the needfor a sufficiently high sampling rate ofthe experimental data to obtain a reason-able resolution of the spatio-temporaldynamics.

• Repetition: experiments need to berepeated to obtain a statistically reliableestimate for the biological variability andother sources of error induced by theexperimental set-up.

The construction of a valid workingmodel of a biological system from experi-mental data can be approached from twodirections:

• Bottom-up modeling: This is based onthe integration of established biologi-cal knowledge on the dynamics of therelevant biological components of thesystem or regulatory network underconsideration. This attempt is usefulwhen most of the reaction partners areknown and their interaction dynamicsare understood. The research goal isthe establishment of an accurate com-puter simulation that allows for: (1)the analysis of the system dynamics;

Systems Biology 15

(2) the scan of parameter ranges that areunattainable in experiments; and (3) theprediction of unknown functionality orinteractions [57]. Attempts at bottom-upmodeling include the λ-phage decisioncircuit [71] or the data-driven simulationof a cancer cell [72].

• Top-down modeling: This approachattempts to apply statistical analysesto data from high-throughput experi-ments derived, for example, from DNAmicroarrays. Data mining techniquessearch for clusters of coexpressed genesas a consequence of cell state or an exter-nal perturbation, such as the knockoutor overexpression of certain genes.

The working assumption for these meth-ods is that coexpressed genes also sharecommon relationships with respect toother biological processes. For example,Muller et al. [73] showed, from clusteringanalysis, how pluripotent stem cells areunder tight control by specific molecularnetworks across species.

The result of a cluster analysis is theconstruction of an interaction network ofgenes or proteins, the topology of whichcan hint at biological reasonable orga-nization principles [33, 74, 75]. Whiletop-down modeling is well in line withthe need for holistic approaches, it hasbeen criticized for violating individual-ity and locality in the cell. Althoughcellular stimulus–response patterns arehighly coordinated, these patterns emergefrom individual protein–protein interac-tions, which cannot be deduced from thehigh-throughput data. Hence, the ques-tion of gaining new biological knowledgeon individual genes or proteins from thisapproach often remains open.

Although knowledge-based bottom-upapproaches of modeling possess acertain appeal to biologists, successful

modeling needs to overcome the gapsin understanding. Most protein–proteininteractions remain unknown, and the un-derlying physics of interacting moleculesrequire better attention. Until now, theinfluence of high molecule concentrationsin the subcompartments of a cell environ-ment on reaction rates is largely unknown,but the spatial distribution of reactantsand their molecular crowding may wellaffect the reactions and their rates [76,77]; an example of this is when explainingthe symmetry breaking process of mitosis[78]. One particular challenge arises whentrying to identify the relevant components,as this is very difficult due to the vast num-ber of combinations of active molecules.Hence, it remains unclear as to howthis approach can be scaled up to largenetworks on the cellular or even tissuelevel. Indeed, this situation presentlyposes major difficulties for detailedmathematical modeling in biology.

Regardless of the approach taken, thereis a need to define the level of model ab-straction, complexity, and spatiotemporalscale of the system under investigation.Cellular processes have characteristic timescales that range from milliseconds forindividual protein–protein interactions tominutes for phosphorylation events, up tohours, days, and years for changes in genetranscription, cell growth, and gene mu-tations, respectively. This has importantconsequences for the appropriate choiceof model detail level. Processes that oc-cur much faster than the time scale ofobservation can be assumed to be in-stantaneous, while slow processes can beassumed to be quasi-static. As a conse-quence, the level of abstraction leads tocertain dynamics being modeled and sim-ulated in detail, whereas for other partsthe details can be neglected – for example,due to low sensitivity towards specific

16 Systems Biology

parameter values. Recently, Busch et al.[79] have used time-scale separation be-tween fast protein signaling and slowtranscription dynamics to infer a dynamicdecision network for hepatocyte growthfactor-induced migration of keratinocytes.On the time-scale of observation, all pro-tein concentrations were assumed to be inquasi-equilibrium, such that it sufficed tofocus on the change in gene regulation asthe decisive element of regulation.

The major goal of theory and modelsimulations is the reverse engineering ofbiological systems, which is ‘‘. . . the pro-cess of analyzing a system to identify itscomponents and their interrelationshipsand create representations of the system inanother form or at a higher level of abstrac-tion’’ [80]. Despite major efforts in thisfield [81], biological systems still pose a ma-jor challenge towards reverse engineering,due to the hidden complexities, inherentrobustness, and possibly suboptimal de-sign of functional units. Circular causalitymakes it difficult to distinguish betweencause and effect from biological data, andall of this imposes ambiguities when at-tempting to deduce the correct biological‘‘wiring diagram’’ from the experimentaldata. It is hoped that systems thinking inbiology, with its concepts of robustness, hi-erarchy and modularity, in addition to thenecessary protocols, will provide detailedbottom-up models with testable hypothe-ses for model discrimination [39].

4.1Network Biology

Network biology is the study of thestatic organization of life as networks ofbiological entities such as DNA, proteins,RNA, and metabolites [33, 82].

The relationship between these entitiesis depicted graphically, the result being a

set of nodes connected by edges. Usually,the nodes represent the state variables ofthe system (such as the molecule con-centrations), while the connections definethe interaction between the nodes. Thenetworks are characterized by their con-nectivity, path length, and clustering distri-butions. Connectivity indicates how manyneighbors each node has on average, whilethe path length denotes the average sep-aration between arbitrarily chosen nodes,and the clustering coefficient is a measureof the grouping tendency of the nodes.

Network biology attempts to discoveruniversal design and organization princi-ples, which govern the functioning andevolution of intercellular and intracel-lular networks [83, 84]. Biological net-works appear to share certain topologiesthat are best described as scale-free net-works; these possess few highly connectednodes, termed hubs, while many nodesshare only a few connections. Hence,the connectivity distribution follows apower-law: P(k) ∝ kn, where k and P(k)denote the connectivity and distribution,respectively, and n is the connectivity ex-ponent. These networks appear if the newnodes have a preferential attachment toalready highly connected hubs [85, 86], al-though, interestingly, this type of complexnetwork also emerges in other aspects oflife and society, such as the worldwide weband social interaction webs [87].

Network architecture and task are closelylinked. By comparing the topology of theregulatory network in E. coli with the callgraph of the kernel of the Linux com-puter operating system, Yan et al. showedthat both networks show a hierarchicallayout, despite having fundamentally dif-ferent design principles [88]. The E. colinetwork is optimized toward robustness,with few global regulators at the top andmany downstream targets, whereas the

Systems Biology 17

Linux kernel is designed for code efficiencyand the re-use of software modules at thecost of robustness, and has many regu-lators controlling a small set of highlyconnected generic functions. By usinghigh-throughput methods, it is possible todraw ever more-detailed interaction websof protein–protein, metabolite, and genetranscription networks. Experimental re-sults obtained with yeast [89] and E. coli[90] seem to support the view that bothmetabolic and genomic regulatory net-works show a hierarchical organizationwith few recurrent subnetworks, termedmotifs, which hint at the existence of el-ementary regulatory units [5, 74]. Whilemodules are discrete functional units thatare semi-detached from the whole sys-tem, motifs comprise a set of genes ormetabolites that form recurring, signif-icant patterns of interconnections, thatare inseparable from the remainder ofthe system [87]. Subsequently, Ravaszet al. showed the metabolic networks of43 distinct organisms to have a modularorganization that was interconnected in ahierarchical manner – a system-level cel-lular organization that might be generic innature [91].

The need to investigate the intercon-nections of genes–proteins and proteins–proteins stems from the fact that thegenome of various species is quite sim-ilar, despite it having been argued that theevolution of ever more-complex speciesis correlated with an increased functionalconnectivity between a constant numberof genes. Motifs might provide a means toincrease the interconnectivity of existingproteins, in this way creating new func-tionalities. As noted by Ravasz et al:

‘‘This is likely one of several rea-sons that the apparent complex-ity of organisms can increase somarkedly without a corresponding

increase in gene number. Anattribute of proteins encoded bythe human genome is that theyhave a richer assembly of domainsthan do their counterparts in in-vertebrates or yeast, and indeedthe assortment of domains intonovel combinations is likely animportant aspect of genome diver-gence’’ [92].

Today, network analysis has openedup new avenues in the global analysis ofdiseases, and their mutual connections.Such analysis allows for the identificationof the genetic basis of disease [93], andcan also reveal novel gene associationoverlapping across common humandisorders, which in turn helps to unravelthe general patterns of human diseasesthat are not clear from studies of theindividual conditions [94].

4.2Dynamic Network Models

Despite the intuitive appeal of network bi-ology for the structure identification ofbiological systems, it is limited in thesense that it does not include the tem-poral dynamics of the system. Life is anemergent property of the interaction ofcellular proteins. Hence, the dynamic in-terplay occurring via chemical reactionsand changing cellular protein numbers isessential to the cell function, and con-stitutes the essence of many modelingapproaches in Systems Biology. Booleannetworks are an abstract, yet relevant, ap-proach towards the inclusion of a temporalevolution of large-scale networks, as theyoffer a qualitative modeling approachesto build and analyze simplified, but stillrigorous, dynamical models. Boolean net-works are used for the elucidation of

18 Systems Biology

large-scale dynamic protein signaling net-works, where it is assumed that the de-tailed inclusion of, perhaps, concentrationgradients or the stochastic effects of pro-tein concentrations, can be neglected infavor of including a large number of play-ers [95]. Each node of a Boolean networkcan assume an ON/OFF value. Then, thestate of each node at the next time step(t + 1) is deduced deterministically froma logical Boolean function (i.e., AND, OR,NAND . . . ), based on its current state andexternal input [96]. Boolean networks havebeen used successfully for the analysisof protein and gene regulatory networks[97], for the modeling of the epidermalgrowth factor receptor (EGFR)/ErbB sig-naling pathways [98], and apoptosis [99].

4.3Reaction–Diffusion Models

Biochemical reactions can be consideredas the most fundamental processes incells, wherein the concentrations of the re-acting species change subject to the othermolecule species involved within the re-spective reactions. Ordinary differentialequations/partial differential equations(ODEs/PDEs) are a natural choice for themathematical description of system dy-namics with continuous system states intime and space, respectively [6, 100]. Theyare, therefore, ideally suited to describechanges in concentrations in biochemi-cal reactions, being the most widespreadformalism to model systems throughoutthe various scientific domains. In biol-ogy, they are widely used to describethe time and space course of molecularconcentrations. In this case, differentialequations relate the rate of change ofa variable (protein substrate) to the cur-rent state of other variables (reactant),wherein the interaction between the

various reactants is modeled throughfunctional and differential relations. Thereis a vast literature available on modelingbiochemical reactions based on differen-tial equations, especially in the contextof metabolic processes (cf. Ref. [100] andreferences therein).

In general, it is possible to distinguishbetween purely time-dependent ODEs,and more general PDEs, which addition-ally include spatial dimensions. A PDEdescribing the spatiotemporal evolution ofa system has the general form:

∂ψi(r, t)

∂t= f

[ψi(r, t)

]︸ ︷︷ ︸

temporal evolution

+∇Di(r, t)∇ψi(r, t),︸ ︷︷ ︸spatial Diffusion

(1)

where ψi(r, t) denotes the respectivesystem state variables of the variousmolecules labeled by the subscript i. Theabove equation reduces to an ODE inthe absence of spatial diffusion, that is,Di(r, t) = 0. The term f [ψ i(r, t)] denotesthe respective synthesis rates, dependingon the various concentrations of ψ i(r, t)and possibly external signals. It is oftencomprised of the various, mostly non-linear, interaction functions between thesystem reactants based on the ‘‘law ofmass action,’’ which leads to, for example,Michaelis–Menten-like enzymatic degra-dation or the Hill-type cooperative activa-tion [6, 100]. The organization of inter-acting molecules within a cell implies theconcept of pathways, in which informa-tion processing in the cell is organized. Interms of mathematics, such a biochemicalnetwork is then represented as a system ofcoupled differential equations, as shownabove. It is those nonlinear interactionsthat are essential for biological systemsto show nontrivial behavior such as mul-tistability, hysteresis, or oscillations [43].

Systems Biology 19

A stability analysis of the system of coupledequations is usually carried out to unravelthe qualitative behavior of steady-state so-lutions and their stability, as well as theoccurrence of periodic solutions in spaceand time [101]. As a matter of fact, the dy-namic behavior of ODE systems dependsheavily on the reaction parameters of theunderlying chemical reactions. Therefore,the estimation and identification of param-eters from experimental data [102, 103],as well as the optimal experiment designto yield a maximal amount of new bi-ological knowledge [104, 105], is a fieldof active research in Systems Biology.Due to the nonlinear interaction termsf [ψ i(r, t)], an analytical solution of the dif-ferential equations is usually impossible,and solutions must be found via numericalintegration. A variety of numerical integra-tion algorithms for ODEs and PDEs can befound in the literature [106]. In addition,various computer software packages havebeen developed specifically for the simula-tion and bifurcation analysis of nonlinearsystems, such as GEPASI [107], DBSolve[108], Cell Designer [109], or a Matlab ex-tension such as SB toolbox [110] or Potterswheel [111].

The great success of modeling is its pre-dictive power. From model simulations, itis possible to obtain a better insight intothe regulatory logic, and it is also possibleto perform experiments in the computerthat otherwise were impossible. A goodexample of this is the mitogen-activatedprotein cascade model, as originally pro-posed by Huang and Ferrell [112]. Here,the pathway is highly conserved and im-plicated in various biological processes,conducting signals from the membrane tothe nucleus. It is composed of three ki-nases that sequentially phosphorylate andactivate each other. By converting and sim-ulating the rate equations into a system of

coupled ODEs, Huang and Ferrell showedthat this particular pathway architecturecould be used to convert a graded re-sponse into a switch-like robust output,appropriate for mediating processes suchas mitogenesis or cell fate induction.

The spatial aspect of the equation entersthe description due to the inclusion of adiffusion term, where the diffusion termDi(r, t) depends on space and time ingeneral. Alan Turing was the first to pointout such reaction–diffusion systems as apossible explanation for morphogenesisin natural systems, when he showed –theoretically – that a system of reactingand diffusing chemicals can evolve sponta-neously to a spatially heterogeneous stateas a response to an infinitesimal smallforcing [113]. These pioneering studiesled to a new branch of research, andmany so-called ‘‘Turing systems’’ havebeen proposed (though not finally proven)to account for pattern formation indevelopmental biology. Examples includean explanation of the pattern formationon snail houses [114], the modeling ofDrosophila embryogenesis [115], or thecoating patterns of animals [100]. Theimportance of spatiotemporal inhomo-geneities in molecular concentrationsand molecular crowding in signal trans-duction pathways has also recently beenacknowledged [77, 116]. As a consequence,cellular functions may also employ activetransport mechanisms to sustain reliablecellular processes [117, 118].

The above ODE formalism fails in thecase of low molecule concentrations [119],as the discrete and random nature of theindividual, elementary reactions betweenmolecules then becomes non-negligible,showing an impact on a macroscopic scaleas fluctuations in the molecule concentra-tion over time [120]. This chemical noisebecomes most significant in the regulation

20 Systems Biology

of gene expression, which is usuallyaccompanied with low copy numbers ofmRNA transcripts and the genes them-selves [121]. Stochastic effects in cellularpathways have been attributed to causephenotypic diversity in isogenetic popula-tions of cells [122], or to play a major rolein the lysis-lysogeny decision circle of theλ-phage [123].

4.4Holism versus Reductionism: The GlobalDynamics of Networks

Currently, most Systems Biology ap-proaches follow either a top-down or abottom-up approach. However, while bothhave their unique motivation, they eachalso hold certain criticisms. Typically, atop-down approach allows for the holistic,unbiased view of cellular events, albeit atthe expense of limiting detail level, thusviolating the individuality and locality thatare important in the understanding of cel-lular processes. The bottom-up approachaims at a detailed mechanistic and causalunderstanding of biochemical networks,but it is not yet been determined howsuch studies of ‘‘isolate’’ signaling path-ways can be scaled up to the cellular oreven tissue and whole organism levels. Asan interim solution, Huang and Ingber[20, 124] have proposed the study of cellu-lar behavior on the level of global networkdynamics to reveal any higher-order, col-lective behavior of the interacting genesand proteins. These authors have de-scribed global network organization interms of a state space, spanned by theexpression levels of the whole genome, aswell as attractors providing a mathemati-cal and molecular basis for an epigeneticlandscape. Genome-wide expression levelscannot take up arbitrary values; instead,they are tightly coupled to respective cell

states – that is, to the various cell types anddistinct, stable phenotypic states in a mul-ticellular organism. Such stable networkconfigurations are referred to as attractorstates. Huang and Ingber have argued thatthe attractors naturally capture the essen-tial properties of cell behavior, such as themutual exclusivity of cell fates, robustness,and all-or-none transitions in response toa large variety of signals. As a proof ofprinciple, it was shown that HL60 cellsare capable of following different routesinto the same differentiated state. Sub-sequently, upon the application of twodifferent stimuli, separate transcriptomeresponse transients were created which,nevertheless, settled onto the same geneexpression pattern after several days [125].

This formal framework on the orches-trated role of cellular gene network dy-namics could serve as a potential expla-nation for stem cell differentiation andthe reprogramming of differentiated backinto pluripotent stem cells [126], or evencancer progression. The accumulation ofgenetic mutations over time would distortthe attractor landscape such that it wouldeventually lead to an altered cellular re-sponse within the same cellular context[124]. While the above suggestions are in-tuitive and appealing, their value must stillbe proven with respect to knowledge gainon the level of individual genes and pro-teins as potential targets for cell controland intervention.

4.5Modeling Resources and Standards

The development of a standardized com-puter infrastructure is of utmost impor-tance to manage the increasing amount ofknowledge relating to biological systems.This supports the effective utilization ofresources and the exchange of models,

Systems Biology 21

ideas, and data. An integrative softwaretool for Systems Biology should comprisethe following features for proper systemsunderstanding [67]. From a mathematicalpoint of view, these tools need to sup-port different simulation algorithms, suchas deterministic and stochastic ODE andPDE solvers, and should include analysisalgorithms for parameter estimation andmodel discrimination.

From an experimental point of view,a package should support a standardizedmodeling language, while in terms ofsoftware the simulation package shouldrun independently from the computerplatform, preferably on a computer gridor cluster environment.

A number of simulation packages arecurrently under active development, mostof them freely available. As a consequenceof the complexity encountered in bio-logical systems, there is no integrativesoftware package presently capable of han-dling all phenomena in signal transduc-tion, metabolic pathways, or spatiotem-poral simulations. Some of the currentlymost versatile packages among many areECell [127], Virtual Cell [128], Cellware[129], Cell Designer [109], or Smart Cell assimulation packages for spatiotemporalsimulation [130].

Model building is a complex enterprise,and is usually accomplished in collabora-tion with scientists from different researchinstitutions. Serving the need for seamlessinformation exchange of computer-basedmodels, progress has been made to definestandardized biological ‘‘wiring diagrams’’[131] together with a common descriptionlanguage, SBGM [132]. Moreover, com-mon standard mark-up languages basedon the XML format help in the unique def-inition of biological entities, pathways, andevents for the rapid exchange of models be-tween experimentalists and theoreticians.

The most common of these are CellML[133], BioPAX [134], and the Systems Bi-ology Markup Language, SBML [135]. TheSystems Biology workbench extends theabove approach by providing a standard-ized application interface for researchersto exchange not only model data andresults, but also the simulation tools them-selves [136]. Another challenge lies in theinherent modeling of the stochasticity ofinner cell processes.

The SSA (stochastic simulation algo-rithm) from Gillespie is computationallyinfeasible with an increasing number ofmolecules, and the computational powerfor exact stochastic simulations will beimmense. Consequently, different algo-rithms have been developed, such as toreduce the number of random variablesfor simulation [137] or to allow larger timesteps to be taken as a justifiable error in therespective reaction probabilities [138]. Sev-eral program packages for the simulationof stochastic molecular dynamics exist, in-cluding StochSim [139] or Stocks [140].

Regardless of the modeling approach ofbiological systems employed, there is al-ways a need for parameter estimates of atleast a few constituents of the model. Be-cause of the sheer amount of parametersrequired for successful modeling, and thehuge number of experiments already per-formed, the need to organize experimentaldata has brought about several publiclyavailable databases of molecular proper-ties, interactions, and pathways [141–143].These provide an invaluable infrastructurefor future modeling efforts, enabling themodeler to begin simulations from a cer-tain degree of abstraction [144].

Spatiotemporal modeling will addition-ally require information concerning thephysical structure of its model con-stituents, as obtained using microscopy.For these needs, the development of an

22 Systems Biology

Open Microscope Environment, which iscurrently being built as a joint effortamong various research institutions, willprovide a unified data format and databaseenvironment for consistently annotating,storing, and retrieving five-dimensionalmicroscope images (four dimensions inspace and time, with additional color in-formation) and exchanging them betweenresearch institutions [145, 146].

5Future Prospects of Systems Biology

The systematic generation and analysis ofquantitative experimental data slowly, butdefinitely, is turning biology into branchof science that is close to engineering.Once the ‘‘language of the genes’’ – thatis, their syntax and their semantics – isdecoded [147], then theoretical knowledgeand experimental expertise will suffice todraft and create synthetic model cells oreven organisms from scratch, on whichnew drugs and cures can be tested in silicobefore their possible assembly in vivo [148].

There is indeed the prospect that Sys-tems Biology will change medical practiceby allowing the prediction of new com-binatorial and/or personalized drugs fordiseases that currently are regarded assevere, including Alzheimer’s, diabetes,human immune deficiency virus (HIV),or cancer [149]. Indeed, Systems Biology isset to change today’s medicine from beingresponsive to being predictive, preventive,personalized, and participatory; this situa-tion is often referred to as ‘‘P4 medicine’’[150]. For example, viewing cancer froma systemic level as a robust system mightprovide physicians with a framework forfuture anticancer strategies [61]. As a con-sequence, anticancer therapies might bedeveloped from mathematical modeling to

first identify the peculiarities of an hetero-geneous tumor cell population, and thentry to predict and control cell activity by de-tecting and using specific fragile points foreach tumor cell type [151]. As a first suc-cess in this line, novel, computationallypredicted anticancer drugs targeting theErbB family of receptor tyrosine kinaseshave been developed from in silico modelsof signaling pathways, and are currentlyundergoing clinical trials [152, 153].

5.1Synthetic Biology

The aim of Synthetic Biology is to(re-)design new or already existing biolog-ical parts, devices, and systems with thehelp of mathematical modeling and engi-neering approaches; in other words, Syn-thetic Biology is the technological counter-part of Systems Biology. Yet, progress inSynthetic Biology goes hand in hand withcurrent progress in cellular and molecularbiology, as well as genetics and associatedfields of engineering and computer sci-ences. It is hoped that the combinationof this knowledge will enable the creationof essentially artificial systems, by employ-ing biological design principles with newcombinations of building modules fromexisting (sub)cellular systems [154, 155].In fact, the first steps in the creation ofartificial genomes and their implantationinto host organisms has already achieveda degree of success [156].

Synthetic Biology departs from acomponent-based approach by viewing aliving system as a programmable entitythat is composed of interacting modules,each having particular functions andwhich exchange their information viaprotocols. The results of current researchhave suggested that these modules are(possibly) limited in number, albeit their

Systems Biology 23

specific tasks are diverse. Synthetic Biol-ogy implies a scientific agenda on a higherlevel of abstraction towards identifyingand categorizing the various module typeswithin the cell and across organisms,and investigating their interactions onthe basis of the modular rather than themolecular interaction. From an engineer-ing point of view, this means that modulescan be taken out of their current evolu-tionary context and assembled differently,in the sense of versatile building blocks.

Ideally, Synthetic Biology starts withmathematically inspired designs, such asa system of coupled differential equationswith desired dynamic properties, whichare then translated into the biological andchemical realities as promoters, enzymes,or metabolites within a cell. Along thisline of thought, the workings of switchesand bistable systems of stabilizing controlloops have been systematically recast interms of chemical reaction schemes [148,157]. Different authors have describedthe fundamental principles of buildinglogic circuits into the language of generegulatory networks. For example, Tysonet al. presented a systematic overviewof designs for biological control systemssuch as switches, sniffers or buzzers, andcombined their mathematical descriptionwith experimental findings [43]. Likewise,Hasty et al. have reviewed the possibilitiesof constructing gene circuits which servevarious functions such as autoregulation,repression, or logical gates [157]. In a firstde novo design study of a gene regulatorynetwork, Guet et al. systematically ana-lyzed the phenotypic behavior of differentparameter and topology combinations in agenetic regulatory network [158]. These au-thors constructed various logic gates, suchas NAND, NOR, and NOT IF, through thecombination of three nodes together withfive promoters. As an important result,

it was found that not only the parametervalues but also the network topology wasimportant in order to determine unam-biguously the computational function ofthe systems.

Recently, efforts have been undertakento establish a Biological InformationSystem (BIS) which extends genomicdatabases with quantitative mechanisticknowledge [70]. The BioBricks foundation(http://bbf.openwetware.org/), foundedby engineers and scientists from theMassachusetts Institute of Technology(MIT), Harvard and the University ofCalifornia, San Francisco (UCSF), haveset up a free, publicly available repositoryof standardized biological parts for thecommon re-use and combination to de-sign de novo functions in living organisms.Bio-Bricks constitute promoters, proteins,RNA-coding sequences, or transcriptionalterminators; physically, they are DNAsequences stored on a circular plasmid dis-tributed by the Registry of Biological Parts(http://www.partsregistry.org). By usingsuch data in an integrated form, it willbecome possible to build refined syntheticsystems by checking on key molecules andreplacing complex pathways with effectivereaction parameters, thus defining biolog-ically meaningful reaction subsets fromthe large amount of possible reactions.

5.2Conclusions: Where Are We?

Although, today, Systems Biology is stillat the start of a new branch of science,the road ahead is clearly marked andcommonly agreed on by many people. Yet,the problems associated with all of theprocesses lie in the details.

To date, most experimental biosciencesare method-driven, and pay muchattention to detail and the production,

24 Systems Biology

in extreme cases, of large amounts ofdata which are out of context. Theoret-ical sciences, on the other hand, areprinciple-driven; they neglect importantdetails and thus come up with theoremsand models that are ‘‘out of this world.’’As a consequence, research groupsmust learn to focus on the importantareas of their respective systems underinvestigation, while also determiningwhich part to measure in detail and whichpart to neglect:

• Biologists need to adopt abstract think-ing, and to trust in the language ofmathematics and necessary simplifica-tion.

• Physicists must learn to abolish over-simplified thinking [159] and becomeused to the analysis of strongly inter-acting systems with many degrees offreedom.

• Engineers must gain a deeper un-derstanding of their systems underinvestigation beyond numerical solu-tions, for example, in terms of processoptimization.

The necessary unification of these vari-ous branches of science will, for the sakeof the advancement of Systems Biology,make communication skills among scien-tists ever more important. Previous yearsof ‘‘isolated’’ research have diversified thescientific vocabulary and methods, makingit sometimes very difficult to mediate anddiscuss interdisciplinary goals and meth-ods. Moreover, all scientists must learnto place their own research into a muchwider frame of interdisciplinary researchand research teams.

Despite this increased complexity ofthe systems under study, one of thegreatest challenges in Systems Biologyis to bridge the gap between detailed

kinetic protein signaling models (ofup to 100 different molecules) andlarge-scale -omics approaches, therebyproviding – simultaneously – data onthousands of genes or proteins. Atpresent, the means by which the detailedmodels can be scaled up to include morevariables and to bridge several time scales(from minutes to hours or even days)is largely unknown. In contrast, it isvery difficult to move from the statisticalanalysis of -omics data to the predictionof individual protein interactions. Yet,taken together, despite not being ableto bridge this modeling gap, there is aclear need for an intuitive understandingof the relevant elements of the systemunder study. Vilar et al. [160] have argued,using the example of the lac operon,that ‘‘. . . even in the ‘postgenomicera,’ (modeling) will still rely more ongood intuition and skills of quantitativebiologists than on the sheer power ofcomputers.’’

Today, it is strongly believed that Sys-tems Biology will promote a unificationof the sciences, especially if the scientificcommunity continues to work ‘‘holisti-cally’’ in a joint effort of biology, math-ematics, physics, and engineering. Onlythen might a new generation of scientistswith an interdisciplinary training emerge,whose everyday business will comprise notonly working in the laboratory but alsoperforming data analysis and model sim-ulations, in addition to exchanging theirdata and results freely through publiclyavailable databases. With the establish-ment of new experimental techniques andmathematical tools, and asking the rightquestions, it is clear that the systems ap-proach to biology will become a majorsuccess. Indeed, it will change not onlymodern life sciences, but also views on lifeitself.

Systems Biology 25

References

1 Kitano, H. (2002) Systems biology: a briefoverview. Science, 295, 1662–1664.

2 Special Issue (2002) Systems biology. Sci-ence, 295.

3 Special Issue (2002) Systems biology. Na-ture Insight, 420.

4 Szallasi, Z., Stelling, J., Periwal, V. (2006)System Modeling in Cell Biology, The MITPress.

5 Alon, U. (2007) An Introduction to SystemsBiology: Design Principles of Biological Cir-cuits, Chapman & Hall/CRC.

6 Klipp, E., Liebermeister, W., Wierling, C.,Kowald, A., Lehrach, H., Herwig, R. (2009)Systems Biology, Wiley-Blackwell.

7 Fu, P., Panke, S. (Eds) (2009) SystemsBiology and Synthetic Biology, AIChE-WileyPress.

8 Smuts, J.C. (1926) Holism and Evolution,Macmillan & Co. Ltd, London.

9 Bertalanffy, L. (1973) General Systems The-ory, Penguin, Harmondsworth.

10 Ashby W.R. (1956) Introduction toCybernetics, Chapman & Hall, London.http://pespmc1.vub.ac.be/books/IntroCyb.pdf (accessed 8 April 2011).

11 Wiener, N. (1948) Cybernetics: Control andCommunications in the Animal and theMachines, The MIT Press, Cambridge.

12 Ideker, T., Galitski, T., Hood, L. (2001)A new approach to decoding life: systemsbiology. Annu. Rev. Genomics Hum. Genet.,2, 343–372.

13 Wolkenhauer, O. (2001) Systems biol-ogy: the reincarnation of systems theoryapplied in biology? Brief. Bioinform., 2 (3),258–270.

14 Mesarovic, M.D. (1968) System Theoryand Biology – View of a Theoretician, in:Mesarovic, M.D. (Ed.) Systems Theory andBiology, Springer, New York, pp. 59–87.

15 Wolkenhauer, O., Hofmeyr, J.-H.S. (2007)An abstract cell model that describes theself-organization of cell function in livingsystems. J. Theor. Biol., 246 (3), 461–476.

16 Westerhoff, H.V., Palsson, B.O. (2004)The evolution of molecular biology intosystems biology. Nat. Biotechnol., 22 (10),1249–1252.

17 Pittendrigh, C.S. (1958) Adaptation, nat-ural selection, and behavior, in: Roe, A.,

Simpson, G.G. (Eds) Behavior and Evolu-tion, Yale University Press, New Haven,pp. 390–416.

18 Baker, M.D., Wolanin, P.M., Stock, J.B.(2006) Signal transduction in bacterialchemotaxis. BioEssays, 28 (1), 9–22.

19 Levine, M., Tjian, R. (2003) Transcriptionregulation and animal diversity. Nature,424, 147–151.

20 Huang, S. (2004) Back to the biologyin systems biology: what can we learnfrom biomolecular networks? Brief. Funct.Genomic. Proteomic., 2 (4), 279–297.

21 Dufva, M. (Ed.) (2009) DNA Microarrays forBiomedical Research, Methods in MolecularBiology, Vol. 529, Humana Press.

22 Hawkins, R.D., Hon, G.C., Ren, B. (2010)Next-generation genomics: an integra-tive approach. Nat. Rev. Genet., 11 (7),476–486.

23 Fields, S., Song, O. (1989) A novel geneticsystem to detect protein-protein interac-tions. Nature, 340 (6230), 245–246.

24 Johnsson, N., Varshavsky, A. (1994) Splitubiquitin as a sensor of protein interactionsin vivo. Proc. Natl Acad. Sci. USA, 91 (22),10340–10344.

25 Spiller, D.G., Wood, C.D., Rand, D.A.,White, M.R.H. (2010) Measurement ofsingle-cell dynamics. Nature, 465 (7299),736–745.

26 Demirel, Y., Sandler, S.I. (2002) Thermody-namics and bioenergetics. Biophys. Chem.,97, 87–111.

27 Hartwell, L.H., Hopfield, J.J., Leibler,S., Murray, A.W. (1999) From molecu-lar to modular cell biology. Nature, 420,C47–C52.

28 Malygin, A.A., Parakhnevitch, N.M.,Ivanov, A.V., Eperon, I.C., Karpova,G.G. (2007) Human ribosomal proteins13 regulates expression of its owngene at the splicing step by a feedbackmechanism. Nucleic Acids Res., 35 (19),6414–6423.

29 Vicsek, T. (2002) Complexity: the biggerpicture. Nature, 418, 131.

30 Carlson, J.M., Doyle, J. (2002) Complexityand robustness. Proc. Natl Acad. Sci. USA,99 (Suppl. 1), 2538–2545.

31 Grassberger, P. (1986) Toward a quanti-tative theory of self-generated complexity.Int. J. Theor. Phys., 25, 907–928.

26 Systems Biology

32 Adami, C. (2002) What is complexity?BioEssays, 24, 1085–1094.

33 Barabasi, A.-L., Oltvai, Z.N. (2004) Networkbiology: understanding the cell’s functionalorganization. Nat. Gen., 5, 101–114.

34 Kitano, H. (2002) Computational systemsbiology. Nature, 420, 206–210.

35 Haken, H. (1987) Advanced Synergetics,Springer, Berlin.

36 Bak, P., Tang, C., Wiesenfeld, K. (1987)Self-organized criticality: an explanationof the 1/f noise. Phys. Rev. Lett., 59,381–384.

37 Carlson, J.M., Doyle, J. (1999) Highlyoptimized tolerance: a mechanism forpower laws in designed systems. Phys. Rev.E, 60 (2), 1412–1427.

38 Csete, M.E., Doyle, J.C. (2002) Reverse en-gineering of biological complexity. Science,295, 1664–1669.

39 Stelling, J., Sauer, U., Szallasi, Z., Doyle,F.J., Doyle, J. (2004) Robustness of cellularfunctions. Cell, 118, 675–685.

40 Kitano, H. (2007) Towards a theory ofbiological robustness. Mol. Syst. Biol., 3,137.

41 Lauffenburger, D.A. (2000) Cell signalingpathways as control modules: complexityfor simplicity? Proc. Natl Acad. Sci. USA,97 (10), 5031–5033.

42 Gu, X. (2003) Evolution of duplicategenes versus genetic robustness againstnull mutations. Trends Genet., 19,354–356.

43 Tyson, J.J., Chen, K.C., Novak, B. (2003)Sniffers, buzzers, toggles and blinkers:dynamics of regulatory and signaling path-ways in the cell. Curr. Opin. Cell Biol., 15,221–231.

44 Novak, B., Tyson, J.J., Gyorffy, B.,Csikasz-Nagy, A. (2007) Irreversible cellcycle transitions are due to systems-levelfeedback. Nat. Cell Biol., 9 (7), 724–728.

45 Lpez-Avils, S., Kapuy, O., Novk, B.,Uhlmann, F. (2009) Irreversibility ofmitotic exit is the consequence ofsystems-level feedback. Nature, 459 (7246),592–595.

46 Clarke, D.J., Gimenez-Abian, J.F. (2000)Checkpoints controlling mitosis. BioEssays,22 (4), 351–363.

47 Kearns, W.G., Liu, J.M. (2001) Cellcycle checkpoint genes and aneuploidy:

a short review. Curr. Genomics, 2 (2),171–180.

48 Fulka, J., Tesarık, J., Loi, P., Moor, R.M.(2000) Manipulating the human embryo:cell cycle checkpoint controls. Cloning, 2(1), 1–7.

49 Barkai, N., Leibler, S. (1997) Robustness insimple biochemical networks. Nature, 387,913–917.

50 Alon, U., Surette, M.G., Barkai, N., Leibler,S. (1999) Robustness in bacterial chemo-taxis. Nature, 397, 168–171.

51 Hartman, J.L., Garvik, B., Hartwell, L.(2001) Principles for the buffering ofgenetic variation. Science, 291, 1001–1004.

52 Gell, M., Noort, V., Yus, E., Chen,W.-H., Leigh-Bell, J., Michalodimitrakis,K., Yamada, T., Arumugam, M., Doerks,T., Khner, S., Rode, M., Suyama, M.,Schmidt, S., Gavin, A.-C., Bork, P., Serrano,L. (2009) Transcriptome complexity in agenome-reduced bacterium. Science, 326(5957), 1268–1271.

53 von Dassow, G., Meir, E., Munro, E.M.,Odell, G. (2000) The segment polaritynetwork is a robust developmental module.Nature, 406, 188–192.

54 Hanahan, D., Weinberg, R.A. (2000)The hallmarks of cancer. Cell, 100 (1),57–70.

55 Yi, T.-M., Huang, Y., Simon, M.I., Doyle,J.C. (2004) Robust perfect adaptation inbacterial chemotaxis through integral feed-back control. Proc. Natl Acad. Sci. USA, 97(9), 4649–4653.

56 Guisbert, E., Yura, T., Rhodius, V.A., Gross,C.A. (2008) Convergence of molecular,modeling, and systems approaches for anunderstanding of the Escherichia coli heatshock response. Microbiol. Mol. Biol. Rev.,72 (3), 545–554.

57 Kitano, H. (Ed.) (2001) Foundations of Sys-tems Biology, The MIT Press, Cambridge,pp. 1–36.

58 Angeli, D., Ferrell, J.E., Sontag, E.D. (2004)Detection of multistability, bifurcations,and hysteresis in a large class of biologicalpositive-feedback systems. Proc. Natl Acad.Sci. USA, 101, 1822–1827.

59 Becker, V., Schilling, M., Bachmann, J.,Baumann, U., Raue, A., Maiwald, T.,Timmer, J., Klingmller, U. (2010) Covering

Systems Biology 27

a broad dynamic range: information pro-cessing at the erythropoietin receptor. Sci-ence, 328 (5984), 1404–1408.

60 Freeman, M. (2000) Feedback control ofintercellular signalling in development.Nature, 408, 313–319.

61 Kitano, H. (2004) Cancer as a robust sys-tem: implications for anticancer therapy.Nat. Rev., 4, 227–235.

62 Carter, G.W., Rush, C.G., Uygun, F.,Sakhanenko, N.A., Galas, D.J., Galitski,T. (2010) A systems-biology approach tomodular genetic complexity. Chaos, 20 (2),026102.

63 Guimera, R., Arenas, A., Dıaz-Guilera,A. (2001) Communication and optimalhierarchical networks. Physica A, 299,247–252.

64 Davidson, E.H., Rast, J.P., Oliveri, P.,Ransick, A., Calestani, C., Yuh, C.-H.,Minokawa, T., Amore, G., Hinman, V.,Arenas-Mena, C., Otim, O., Brown, C.T.,Livi, C.B., Lee, P.Y., Revilla, R., Rust,A.G., Pan, Z., Schilstra, M.J., Clarke,P.J.C., Arnone, M.I., Rowen, L., Cameron,R.A., McClay, D.R., Hood, L., Bolouri, H.(2002) A genomic regulatory network fordevelopment. Science, 295, 1669–1678.

65 Laub, M.T., McAdams, H.H., Feldblum,T., Fraser, C.M., Shapiro, L. (2000) Globalanalysis of the genetic network control-ling a bacterial cell cycle. Science, 290,2144–2148.

66 Pearl, J. (2000) Causality, Cambridge Uni-versity Press.

67 Dhar, P.K., Zhu, H., Mishra, S.K. (2004)Computational approach to systems bi-ology: from fraction to integration andbeyond. IEEE Trans. Nanobiosci., 3 (3),144–152.

68 Crampin, E.J., Schnell, S., McSharry, P.E.(2004) Mathematical and computationaltechniques to deduce complex biochemicalreaction mechanisms. Prog. Biophys. Mol.Biol., 86 (1), 77–112.

69 Ideker, T., Thorsson, V., Ranish, J.A.,Christmas, R., Buhler, J., Eng, J.K.,Bumgarner, R., Goodlett, D.R., Aebersold,R., Hood, L. (2001) Integrated genomicand proteomic analyses of a systematicallyperturbed metabolic network. Science, 292,929–934.

70 Endy, D., Brent, R. (2001) Modelling cellu-lar behavior. Nature, 409, 391–395.

71 McAdams, H., Shapiro, L. (1995) Circuitsimulations of genetic networks. Science,269, 650–656.

72 Christopher, R., Dhiman, A., Fox, J.,Gendelman, R., Haberichter, T., Kagle,D., Spizz, G., Khalil, I.G., Hill, C. (2004)Data-driven computer simulation of hu-man cancer cell. Ann. N. Y. Acad. Sci., 1020,132–153.

73 Muller, F.-J., Laurent, L.C., Kostka, D.,Ulitsky, I., Williams, R., Lu, C., Park,I.-H., Rao, M.S., Shamir, R., Schwartz,P.H., Schmidt, N.O., Loring, J.F. (2008)Regulatory networks define phenotypicclasses of human stem cell lines. Nature,455 (7211), 401–405.

74 Jeong, H., Tombor, B., Albert, R., Oltvai,Z.N., Barabasi, A.-L. (2000) The large-scaleorganization of metabolic networks. Na-ture, 407, 651–654.

75 Jeong, H., Mason, S.P., Barabasi, A.-L.,Oltvai, Z.N. (2001) Lethality and centralityin protein networks. Nature, 411, 41–42.

76 Elowitz, M.B., Surette, M.G., Wolf, P.-E.,Stock, J.B., Leibler, S.J. (1999) Proteinmobility in the cytoplasm of Escherichiacoli. J. Bacteriol., 181, 197–203.

77 Schnell, S., Turner, T.E. (2004) Reac-tion kinetics in intracellular environmentswith macromolecular crowding: simula-tions and rate laws. Prog. Biophys. Mol.Biol., 85, 235–260.

78 Howard, M., Rutenberg, A.D. (2003) Pat-tern formation inside bacteria: fluctuationsdue to the low copy number of proteins.Phys. Rev. Lett., 90 (1–4), 128102.

79 Busch, H., Camacho, D., Rogon, Z.,Breuhahn, K., Angel, P., Eils, R.,Szabowski, A. (2008) Gene networkdynamics controlling keratinocytemigration. Mol. Syst. Biol., 4, 199.

80 Chikofsky, E.J., Cross, J.H. (1990) Reverseengineering and design recovery: A taxon-omy. IEEE Softw., 7, 13–17.

81 He, F., Balling, R., Zeng, A.-P. (2009)Reverse engineering and verification ofgene networks: principles, assumptions,and limitations of present methods andfuture perspectives. J. Biotechnol., 144 (3),190–203.

82 Strogatz, S.H. (2001) Exploring complexnetworks. Nature, 410, 268–276.

83 Sharom, J.R., Bellows, D.S., Tyers, M.(2004) From large networks to small

28 Systems Biology

molecules. Curr. Opin. Chem. Biol., 8,81–90.

84 Alm, E., Arkin, A.P. (2003) Biologicalnetworks. Curr. Opin. Struct. Biol., 13,193–202.

85 Barabasi, A.-L., Albert, R. (1999) Emer-gence of scaling in random networks. Sci-ence, 286, 509–512.

86 Keller, E.F. (2005) Revisiting ‘‘scale-free’’networks. BioEssays, 27 (10), 1060–1068.

87 Milo, R., Shen-Orr, S., Itzkovitz, S.,Kashtan, N., Chklovskii, D., Alon, U. (2002)Network motifs: Simple building blocks ofcomplex networks. Science, 298, 824–827.

88 Yan, K.K., Fang, G., Bhardwaj, N., R.P.,Alexander, Gerstein, M. (2010) Comparinggenomes to computer operating systemsin terms of the topology and evolution oftheir regulatory control networks. Proc. NatlAcad. Sci. USA, 107 (20), 9186–9191.

89 Lee, T.I., Rinaldi, N.J., Robert, F.,Odom, D.T., Bar-Joseph, Z., Gerber,G.K., Hannett, N.M., Harbison, C.T.,Thompson, C.M., Simon, I., Zeitlinger,J., Jennings, E.G., Murray, H.L., Gordon,D.B., Ren, B., Wyrick, J.J., Tagne, J.-B.,Volkert, T.L., Fraenkel, E., Gifford,D.K., Young, R.A. (2002) Transcriptionalregulatory networks in Saccharomycescerevisiae. Science, 298, 799–804.

90 Almaas, E., Kovacs, B., Vicsek, T., Oltvai,Z.N., Barabasi, A.-L. (2004) Global organi-zation of metabolic fluxes in the bacteriumEscherichia coli. Nature, 427, 839–843.

91 Ravasz, E., Somera, A.L., Mongru, D.A.,Oltvai, N., Barabasi, A.-L. (2002) Hierarchi-cal organization of modularity in metabolicnetworks. Science, 297, 1551–1555.

92 Pawson, T., Nash, P. (2003) Assembly ofcell regulatory systems through protein in-teraction domains. Science, 300, 445–452.

93 de la Fuente, A. (2010) From differen-tial expression to differential networkingidentification of dysfunctional regulatorynetworks in diseases. Trends Genet., 26,326–333.

94 Goh, K.-I., Cusick, M.E., Valle, D., Childs,B., Vidal, M., Barabsi, A.-L. (2007) Thehuman disease network. Proc. Natl Acad.Sci. USA, 104 (21), 8685–8690.

95 Saez-Rodriguez, J., Alexopoulos, L.G.,Epperlein, J., Samaga, R., Lauffenburger,D.A., Klamt, S., Sorger, P.K. (2009)Discrete logic modeling as a means to

link protein signalling networks withfunctional analysis of mammalian signaltransduction. Mol. Syst. Biol., 5, 331.

96 Jong, H. (2002) Modeling and simulationof generic regulatory systems: a literaturereview. J. Comput. Biol., 9 (1), 67–103.

97 Faur, A., Thieffry, D. (2009) Logical mod-elling of cell cycle control in eukaryotes:a comparative study. Mol. Biosyst., 5 (12),1569–1581.

98 Samaga, R., Saez-Rodriguez, J.,Alexopoulos, L.G., Sorger, P.K., Klamt, S.(2009) The logic of egfr/erbb signaling:theoretical properties and analysis ofhigh-throughput data. PLoS Comput. Biol.,5 (8), e1000438.

99 Schlatter, R., Schmich, K., Vizcarra, I.A.,Scheurich, P., Sauter, T., Borner, C.,Ederer, M., Merfort, I., Sawodny, O. (2009)On/off and beyond – a Boolean modelof apoptosis. PLoS Comput. Biol., 5 (12),e1000595.

100 Murray, J.D. (1993) Mathematical Biology,Springer, Heidelberg.

101 Cross, M.C., Hohenberg, P.C. (1993) Pat-tern formation outside of equilibrium. Rev.Mod. Phys., 65 (3), 851–1112.

102 Moles, C.G., Mendes, P., Banga, J.R. (2003)Parameter estimation in biochemical path-ways: A comparison of global optimizationmethods. Genome Res., 13, 2467–2474.

103 Hengl, S., Kreutz, C., Timmer, J., Maiwald,T. (2007) Data-based identifiability analysisof non-linear dynamical models. Bioinfor-matics, 23, 2612–2618.

104 Bandara, S., Schlder, J.P., Eils, R., Bock,H.G., Meyer, T. (2009) Optimal experimen-tal design for parameter estimation of a cellsignaling model. PLoS Comput. Biol., 5 (11),e1000558.

105 Skanda, D., Lebiedz, D. (2010) An optimalexperimental design approach to modeldiscrimination in dynamic biochemicalsystems. Bioinformatics, 26 (7), 939–945.

106 Press, W.H., Teukolsky, S.A., Vetterling,W.T., Flannery, B.P. (1993) NumericalRecipes in C, 2nd edn, Cambridge Uni-versity Press, New York.

107 Mendes, P. (1993) Gepasi: a softwarepackage for modelling the dynamics, steadystates and control of biochemical and othersystems. CABIOS, 9 (5), 563–571.

108 Goryanin, I., Hodgman, T.C., Selkov,E. (1999) Mathematical simulation and

Systems Biology 29

analysis of cellular metabolism andregulation. Bioinformatics, 15 (9), 749–758.

109 Funahashi, A., Matsuoka, Y., Jouraku,A., Morohashi, M., Kikuchi, N., Kitano,H. (2008) Celldesigner 3.5: a versatilemodeling tool for biochemical networks.Proc. IEEE, 96 (8), 1254–1265.

110 Schmidt, H., Jirstrand, M. (2006) Systemsbiology toolbox for matlab: a computationalplatform for research in systems biology.Bioinformatics, 22 (4), 514–515.

111 Maiwald, T., Timmer, J. (2008) Dynami-cal modeling and multi-experiment fittingwith Potters wheel. Bioinformatics, 24 (18),2037–2043.

112 Huang, C.Y., Ferrell, J.E. (1996) Ultrasen-sitivity in the mitogen-activated proteinkinase cascade. Proc. Natl Acad. Sci. USA,93 (19), 10078–10083.

113 Turing, A.M. (1952) The chemical basis ofmorphogenesis. Phil. Trans. R. Soc. LondonB, 237, 37–72.

114 Meinhardt, H. (2008) Models of biologicalpattern formation: from elementary stepsto the organization of embryonic axes. Curr.Top. Dev. Biol., 81, 1–63.

115 Kauffman, S.A. (1993) The Origins of Order:Self-Organization and Selection in Evolution,Oxford University Press, New York.

116 Volz, D., Eigel, M., Athale, C., Bastian,P., Hermann, H., Kappel, C., Eils, R.(2005) Spatial Modeling and Simulation ofDiffusion in Nuclei of Living Cells, CMSB2004, Lecture Notes in Computer Science,Springer, Heidelberg, pp. 161–171.

117 Kholodenko, B.N. (2003) Four-dimensionalorganization of protein kinase signalingcascades: the roles of diffusion, endocytosisand molecular motors. J. Exp. Biol., 206,2073–2082.

118 Slepchenko, B.M., Schaff, J.C., Carson,J.H., Loew, L.M. (2002) Computationalcell biology: Spatiotemporal simulation ofcellular events. Ann. Rev. Biophys. Biomol.Struct., 31, 423–441.

119 van Kampen, N.G. (1992) Stochastic Pro-cesses in Physics and Chemistry, ElsevierScience Publishers B.V., Amsterdam.

120 Kaern, M., Elston, T.C., Blake, W.J., Collins,J.J. (2005) Stochasticity in gene expression:from theories to phenotypes. Nat. Rev.Genet., 6 (6), 451–464.

121 Golding, I., Paulsson, J., Zawilski, S.M.,Cox, E.C. (2005) Real-time kinetics of gene

activity in individual bacteria. Cell, 123 (6),1025–1036.

122 Elowitz, M., Levine, A., Siggla, W., Swain,P. (2002) Stochastic gene expression in asingle cell. Science, 297, 1183–1186.

123 Arkin, A., Ross, J., McAdams, H.H.(1998) Stochastic kinetic analysis of de-velopmental pathway bifurcation in phagelambda-infected Escherichia coli cells. Ge-netics, 149 (4), 1633–1648.

124 Huang, S., Ingber, D.E. (2007) Anon-genetic basis for cancer progressionand metastasis: self-organizing attractorsin cell regulatory networks. Breast Dis., 26,27–54.

125 Huang, S., Eichler, G., Bar-Yam, Y., Ingber,D.E. (2005) Cell fates as high dimensionalattractor states of a complex gene regulatorynetwork. Phys. Rev. Lett., 94 (12), 128701.

126 Huang, S. (2009) Reprogramming cellfates: reconciling rarity with robustness.BioEssays, 31 (5), 546–560.

127 Tomita, M., Hashimoto, K., Takahashi, K.,Shimizu, T., Matsuzaki, Y., Miyoshi, F.,Saito, K., Tanida, S., Yugi, K., Venter,J.C., Hutchison, C.M. (1999) Ecell: softwareenvironment for whole cell simulation.Bioinformatics, 15 (5), 72–74.

128 Slepchenko, B.M., Schaff, J.C., Macara,I.G., Loew, L.M. (2003) Quantitative cellbiology with the virtual cell. Trends CellBiol., 13, 570–576.

129 Dhar, P., Meng, T.C., Somani, S., Ye, L.,Sairam, A., Chitre, M., Hao, Z., Sakharkar,K. (2004) Cellware – multi-algorithmic soft-ware for computational systems biology.Bioinformatics, 20, 1319–1321.

130 Ander, M., Beltrao, P., Ventura, B.D.,Ferkinghoff-Borg, J., Foglierini, M.,Kaplan, A., Lemerle, C., Tomas-Oliveira,I., Serrano, L. (2004) SmartCell, aframework to simulate cellular processesthat combines stochastic approximationwith diffusion and localisation: analysis ofsimple networks. Syst. Biol., 1, 129–138.

131 Kitano, H., Funahashi, A., Matsuoka, Y.,Oda, K. (2005) Using process diagrams forthe graphical representation of biologicalnetworks. Nat. Biotechnol., 23 (8), 961–966.

132 Novre, N.L., Hucka, M., Mi, H., Moodie,S., Schreiber, F., Sorokin, A., Demir, E.,Wegner, K., Aladjem, M.I., Wimalaratne,S.M., Bergman, F.T., Gauges, R., Ghazal,P., Kawaji, H., Li, L., Matsuoka, Y.,

30 Systems Biology

Villger, A., Boyd, S.E., Calzone, L.,Courtot, M., Dogrusoz, U., Freeman, T.C.,Funahashi, A., Ghosh, S., Jouraku, A.,Kim, S., Kolpakov, F., Luna, A., Sahle,S., Schmidt, E., Watterson, S., Wu, G.,Goryanin, I., Kell, D.B., Sander, C., Sauro,H., Snoep, J.L., Kohn, K., Kitano, H. (2009)The systems biology graphical notation.Nat. Biotechnol., 27 (8), 735–741.

133 Lloyd, C.M., Halstead, M.D.B., Nielsen,P.F. (2004) CellML: its future, present andpast. Prog. Biophys. Mol. Biol., 85, 433–450.

134 BioPAX Homepage Internet (2010),http://www.biopax.org/ (accessed 8 April2011).

135 Hucka, M., Finney, A., Sauro, H.M.,Bolouri, H., Doyle, J.C., Kitano, H.,Arkin, A.P., Bornstein, B.J., Bray, D.,Cornish-Bowden, A., Cuellar, A.A.,Dronov, S., Gilles, E.D., Ginkel, M.,Gor, V., Goryanin, I.I., Hedley, W.J.,Hodgman, T.C., Hofmeyr, J.-H., Hunter,P.J., Juty, N.S., Kasberger, J.L., Kremling,A., Kummer, U., Le Novre, N., Loew,L.M., Lucio, D., Mendes, P., Minch,E., Mjolsness, E.D., Nakayama, Y.,Nelson, M.R., Nielsen, P.F., Sakurada, T.,Schaff, J.C., Shapiro, B.E., Shimizu, T.S.,Spence, H.D., Stelling, J., Takahashi, K.,Tomita, M., Wagner, J., Wang, J. (2003)The systems biology markup language(SBML): a medium for representation andexchange of biochemical network models.Bioinformatics, 19, 524–531.

136 Hucka, M., Finney, A., Sauro, H.M.,Bolouri, H., Doyle, J., Kitano, H. (2002)The erato systems biology workbench: en-abling interaction and exchange betweensoftware tools for computational biology.Pac. Symp. Biocomput., 450–461.

137 Gibson, M.A., Bruck, J. (2000) Efficientexact stochastic simulation of chem-ical systems with many species andmany channels. J. Phys. Chem., 104,1876–1889.

138 Gillespie, D.T. (2001) Approximate accel-erated stochastic simulation of chemicallyreacting systems. J. Chem. Phys., 115,1716–1733.

139 Morton-Firth, C.J. (1998) Stochastic sim-ulation of cell signalling pathways. PhDthesis, University of Cambridge.

140 Kierzek, A.M. (2002) Stocks: stochastickinetic simulations of biochemical systems

with Gillespie algorithm. Bioinformatics, 18(3), 470–481.

141 Lindvall, J.M., Emelie, K., Blomberg, M.,Smith, C.I.E. (2003) In silico tools for signaltransduction research. Brief. Bioinform., 4,315–324.

142 Jensen, L.J., Kuhn, M., Stark, M., Chaffron,S., Creevey, C., Muller, J., Doerks, T.,Julien, P., Roth, A., Simonovic, M., Bork,P., Mering, C. (2009) String 8 – a globalview on proteins and their functionalinteractions in 630 organisms. Nucleic AcidsRes., 37 (Database issue), D412–D416.

143 Kanehisa, M., Goto, S., Furumichi, M.,Tanabe, M., Hirakawa, M. (2010) Kegg forrepresentation and analysis of molecularnetworks involving diseases and drugs.Nucleic Acids Res., 38 (Database issue),D355–D360.

144 Arkin, A.P. (2001) Synthetic cell biology.Curr. Opin. Biotechnol., 12 (6), 638–644.

145 Swedlow, J.R., Goldberg, I., Brauner, E.,Sorger, P.K. (2003) Informatics and quan-titative analysis in biological imaging. Sci-ence, 300, 100–102.

146 Goldberg, I., Allan, C., Burel, J.-M.,Creager, D., Falconi, A., Hochheiser, H.,Johnston, J., Mellen, J., Sorger, P.K.,Swedlow, J.R. (2005) The Open MicroscopyEnvironment (OME) Data Model and XMLFile: Open Tools for Informatics and Quan-titative Analysis in Biological Imaging.Genome Biol., 6, R47.

147 Searls, D.B. (2002) The language of thegenes. Nature, 420, 211–217.

148 Francois, P., Hakim, V. (2004) Design ofgenetic networks with specified functionsby evolution in silico. Proc. Natl Acad. Sci.USA, 101 (2), 580–585.

149 Butcher, E.C., Berg, E.L., Kunkel, E.J. (2004)Systems biology in drug discovery. Nat.Biotechnol., 22 (10), 1253–1259.

150 Hood, L. (2008) A personal journey of dis-covery: developing technology and chang-ing biology. Annu. Rev. Anal. Chem. (PaloAlto CA), 1, 1–43.

151 Fitzgerald, J.B., Schoeberl, B., Nielsen,U.B., Sorger, P.K. (2006) Systems biologyand combination therapy in the quest forclinical efficacy. Nat. Chem. Biol., 2 (9),458–466.

152 Schoeberl, B., Pace, E.A., Fitzgerald, J.B.,Harms, B.D., Xu, L., Nie, L., Linggi, B.,Kalra, A., Paragas, V., Bukhalid, R.,

Systems Biology 31

Grantcharova, V., Kohli, N., West, K.A.,Leszczyniecka, M., Feldhaus, M.J., Kudla,A.J., Nielsen, U.B. (2009) Therapeuti-cally targeting erbb3: a key node inligand-induced activation of the erbbreceptor-pi3k axis. Sci. Signal., 2 (77),ra31.

153 Hendriks, B.S. (2010) Functional pathwaypharmacology: chemical tools, pathwayknowledge and mechanistic model-basedinterpretation of experimental data. Curr.Opin. Chem. Biol., 14 (4), 489–497.

154 Purnick, P.E.M., Weiss, R. (2009) Thesecond wave of synthetic biology: frommodules to systems. Nat. Rev. Mol. Cell.Biol., 10 (6), 410–422.

155 Kiel, C., Yus, E., Serrano, L. (2010) Engi-neering signal transduction pathways. Cell,140 (1), 33–47.

156 Gibson, D.G., Glass, J.I., Lartigue, C.,Noskov, V.N., Chuang, R.-Y., Algire, M.A.,Benders, G.A., Montague, M.G., Ma, L.,

Moodie, M.M., Merryman, C., Vashee, S.,Krishnakumar, R., Assad-Garcia, N.,Andrews-Pfannkoch, C., Denisova, E.A.,Young, L., Qi, Z.-Q., Segall-Shapiro, T.H.,Calvey, C.H., Parmar, P.P., Hutchison,C.A., Smith, H.O., Venter, J.C. (2010)Creation of a bacterial cell controlled by achemically synthesized genome. Science,329 (5987), 52–56.

157 Hasty, J., McMillen, D., Collins, J.J. (2002)Engineered gene circuits. Nature, 420,224–230.

158 Guet, C.C., Elowitz, M.B., Hsing, W.,Leibler, S. (2002) Combinatorial synthe-sis of genetic networks. Science, 296,1466–1470.

159 May, R.M. (2004) Uses and abuses of math-ematics in biology. Science, 303, 790–793.

160 Vilar, J.M.G., Kueh, H.Y., Barkai, N.,Leibler, S. (2002) Mechanisms of noise-resistance in genetic oscillators. Proc. NatlAcad. Sci. USA, 99, 5988–5992.