2.1.1 Systems biology

The simultaneous measurement of the majority of molecular agents (metabolites, proteins, mRNA) in cells is nowadays possible thanks to the development of a series of high-throughput techniques.

Consequently, complete datasets that define very precisely the state and composition of cells under specific conditions are available for study.

Even more than information about the particular elements,
interactions (enzyme-substrate or protein-protein interactions) among those are already becoming well known.

Systems biology (Kitano 2002; Klipp et al. 2005; Palsson 2006) is more aimed to the study of the interactions or links that connect the cellular elements than the elements themselves.

Those links spread across the cell connecting all the components and producing a system extremely complex and highly interconnected.

Being able to connect, integrate and use all those networks to obtain useful biological information is the fundamental promise of systems biology.

From molecules to systems

The actual etymology of the discipline sheds light to the two fundamental roots in its development: molecular biology and systemic or integrative thinking.

Molecular systems biology as it is known now appeared as a viable and promising approach only a couple of decades ago with the proliferation of high-throughput technologies (first the massive parallel sequencing and the other soon after).

The second half of the 20th century saw a series of ground-breaking discoveries that pushed biology into the molecular level: structure of DNA and proteins, the restriction enzymes and cloning.

These advances established the biotechnology industry as it is known now.

After that, the size and scope of some experimental techniques grew (e.g. PCR) and finally automated DNA sequencing techniques took the world by storm in the mid-1990s reaching the present ’omics’ era.

This led to the appearance of bioinformatics in order to extract useful information from the massive amount of data that it was being produced.

Soon after that, a more nominal approach to the analysis, integration and modelling of this data appeared originating the molecular systems biology discipline.

Non-equilibrium thermodynamics can be considered the very beginning of the study of integrated processes in living systems (Westerhoff and Palsson 2004).

From then on most of the system analysis dealt with the bioenergetics of the cell and coupling process principles (Mitchell 1961, 1966).

Subsequently, metabolic models developed quickly in the last decades of the past century, either pathwaylevel kinetic (a brief introduction to kinetic modelling in biology and stochastic simulations can be found in Resat, Petzold, and Pettigrew 2009) or genome-scale constraint-based (Edwards and Palsson 1999).

Probably the most notorious attempt to create a theoretical system framework for biological entities was the general system theory (Bertalanffy 1968).

Later on metabolic control analysis (see a general introduction in Fell 1992) and the close biochemical system theory (Savageau 1969) were noteworthy attempts of characterizing properties of networks of chemical reactions.

These two methodologies showed that some of the properties of the molecular components escape the traditional reductionist approach.

They depend on the structure of the network and therefore a network-wise kind of analysis is needed.

A vast range of approaches are now available for researchers to deal with biological data: kinetic (stochastic and deterministic) and constraint-based modelling, network and graph analysis, multivariate statistics and data fusion.

The need for integrative thinking was always important in biological research.

Nowadays, the raw amount of data we are able to produce makes this need stronger than ever.

Ideally, both sides of a system approach in biology should feed back each other: experimental research fuelling modelling and mathematical description pointing experimental work in the most promising and useful directions.

Components vs. systems

Components come and go, however the system remains (Palsson 2015).

As it has been mentioned, high-throughput technologies completely changed the landscape of biological research.

Information was suddenly available not for one gene or protein but for thousands a the same time.

However, this shift was not merely a scale-up of the subjects of study.

This required a new approach much more suitable for large amounts of data.

Viewing components as part of a system and study them as such is the core idea behind the disciple of systems biology.

Although traditional molecular biology research is still very potent and necessary, the systemic approach for solving multitude of biological problems is simply a necessity.

Phenotype is the result of a large and very complex combination of elements and therefore to study the genotype-phenotype relationship an integrative approach is usually very useful.

This approach is able to highlight the systemic or emergent properties of the system.

These properties arise from the structure of the system itself and escape the traditional, reductionistic and more isolated approach.

Figure 2.1 shows a simple diagram connecting those concepts.

Top-down vs Bottom-up

Metabolic networks are located at the heart of the present systems biology research.

They have been very well studied for many decades and there is a lot of information available to develop and contrast models.

When studying metabolic networks one faces a methodology crossroad that actually spreads to any type of molecular network and that defines the two main research streams in systems biology.

This dichotomy is usually known as the top-down and the bottom-up approaches (Bruggeman and Westerhoff 2007).

Figure 2.1. Different approaches and data sources of traditional molecular biology and systems biology.

The top-down approach starts with cell-wide experimental data and aims to unravel and characterize new molecular mechanisms closer to the components and their interactions.

Therefore it goes from top (genome, proteome, transcriptome data) to bottom (genes, proteins interactions, metabolic reactions).

The work-flow starts with the experimental data, which is analysed and integrated to find out correlations between particular elements and concludes with testable hypothesis about the behaviour and relationship among those components.

Methodologically it uses inference techniques to build phenomenological models (not based on known mechanisms and biological knowledge) to reverse-engineering from solely cell data.

On the other hand, the bottom-up approach tries to elucidate the functions and states of a subsystem that has been studied and characterized with detail.

It begins with the components or constitutive parts (bottom) and then formulates and uses the interactions (usually reactions) among those elements to predict the system behaviour.

The models constructed from this perspective are mechanistic rather than purely phenomenological.

Characteristics

Systems biology aims too understand, explain and characterize biological entities.

Its ultimate objective is no different from traditional molecular biology or even other branches of the biological sciences like ecology or zoology.

It is the system approach to achieve those goals that sets it apart.

In the near future this integrative approach will be so embedded in the stream of research that the "system" label will probably disappear and systems biology will be call just biology.

Figure 2.2. Traditional workflow in systems biology.

Systems biology is a inter-disciplinary field with basically three main legs: biological objects of study, experimental techniques and purpose, data gathering, integration and fusion and mathematical modelling and description.

The stream from one leg to another is pretty straight forward (Figure 2.2).

At the beginning, there is a particular biological system (in this usually a cell or a cellular subsystem) of interest.

Experimental techniques available nowadays allow us to generate massive amounts of data for that particular subsystem.

This data is the integrated if it comes from different techniques or if it partially overlaps with existing data in the literature.

Then, mathematical tools are used to construct a model that explains the data gathered and that is able to make useful predictions (Klipp et al. 2005).

Finally, these predictions are tested in the laboratory and the process can start again (refining or scaling-up the model or changing the subsystem to study).

Applications and challenges

metabolic functions (Guzmán et al. 2015), dynamic estimation of metabolic fluxes (Vercammen, Logist, and Impe 2014), industrial biotechnology (González-Martínez et al. 2014), gene similarity (Fuxman Bass et al. 2013), biological optimality (Schuetz et al. 2012) and even bacterial computing (Amos et al. 2015).

The system approach will become another available toolbox for biological research as molecular biology did at the end of the last century.

In its fast growth systems biology faces important challenges.

The way we manage to confront these challenges will surely determine the shape of the biological research in this century.

First and maybe most importantly is the connection between experiments and modelling.

Still much needs to be done to close the circle of the Figure 2.
2 and generate a real, stable and fluid feedback between data gathering and mathematical modelling.

Secondly, data integration is probably the biggest bottleneck in systems biology right now.

The ability to find, use and share biological models, networks and equations needs to reach the level of coherence found in genome sequences or protein structure.

For this, scientific communication and database development is of the utmost importance.

And finally, we need to start developing models that integrate different biological sources of information.

Metabolite and protein concentrations, expression data, genome sequence must come together eventually to achieve a true whole-cell model.

Particular fields of application such as medicine, drug development, food production, industrial biotechnology, environmental sustainability benefit already from a more systemic approach of molecular biology.

This trend will continue to develop along with other promising sides of the multi-layered biological research.