2.2.1 Networks

Why is the network the most successful representation of large biological systems? And why is it the perfect fit for a systemic approach of biology? The answer to those questions is multiple.

Networks are abstractions of much more complex systems and they emphasises the interactions between elements more than the nature or characteristics of each isolated element (Klipp et al. 2011).

This produces a series of advantages that prove to be useful.

Networks are a good first step into systems for which there is not a lot of information available.

They are very understandable when compare to some mathematical descriptions.

Network topology is a quick way to compare whether two systems are similar.

Furthermore, when compare to random networks, structural features appear highlighting the most informative sections of the network.

Moreover, some processes are especially more suited for a network-type description (disease spread, food chains, neural systems) than a more traditional reductionist approach.

Finally, when dealing with massive amounts of data (i.e. high-throughput technologies) networks may point to potential functions of particular branches.

It is the same idea behind the famous principle that structure defines function in molecular biology only that in a network scale: network structure defines network function.

Biological networks can be divided according to scale in macroscopic and microscopic (Junker 2008).

Macroscopic networks

The very beginning of network analysis and representation in biology is found in ecology.

It is defined as the science that studies the interactions between organisms and their environment.

The interactions between different species are known as ecological networks.

Probably the most famous example of this type of network is food webs, especially predator-prey.

These describe who is affected and in which way in a feeding interaction.

This knowledge is very important for understanding the mechanisms that govern populations and entire ecosystems.

Other common ecological network is plant-pollinator interaction network.

The second most frequent example of macroscopic network is phylogenetic networks.

These represent the evolutionary relationships between organisms.

Usually, these relations were illustrated by tress where branching points represented the of two species during evolution.

However, recent discoveries such as multi-separation and reticulate relationships picture a much more complex landscape.

Microscopic networks

Microscopic or biochemical networks have been studied for many years.

Traditional molecular biology approaches focused mainly on the particularities of each component: activity of one enzyme, targets of a transcription factor, interactions of one protein.

The process was a solid but slow build-up from components to higher order structures such as pathways and complete networks.

High-throughput technologies massively increased the scope and speed of network-based molecular biology (Covert et al. 2004).

Gene regulatory networks, metabolic networks and protein interaction networks are the basic three types of biochemical networks.

Transcription regulatory networks control which genes are express in each moment in the cell (Carrera, Elena, and Jaramillo 2012; Carrera et al. 2009; Even, Lindley, and Cocaign-Bousquet 2003; Perrenoud and Sauer 2005; Thiele et al. 2009).

The expression of one gene is controlled by the product (usually proteins called transcription factors) of another.

Gaining insight into how the cells controls the expression of its genes depending on internal or external conditions is the next big step in systems biology.

Small motif such as feed-forward and feed-back loops are very common in these networks.

On the other hand, signalling networks are a particular kind of regulatory networks that contain signalling cascades or chains usually associated with phosphorilation events.

These networks link intracellular processes to extracellular environments and adjust cellular functions according to those external conditions.

Proteins operate mostly interacting with almost all kinds of molecules in the cell: small metabolites, lipids, nucleic acids or other proteins.

Protein interaction networks illustrate how proteins associate with other proteins to carry out their functions (De Las Rivas and Fontanillo 2010; Fossum et al. 2009; Rajagopala et al. 2014).

There are several experimental and computational methods to determine these networks.

Metabolic networks display metabolites being transformed into each other due to the presence of enzymes.

Metabolic networks are by far the most studied biochemical network (Chassagnole et al.

2002; Link, Christodoulou, and Sauer 2014; Vercammen, Logist, and Impe 2014).

Biochemists have been studying metabolic reactions and linking them with other reactions for many decades.

A lot of metabolic constraint-based static models and pathway-level dynamic models have been developed in the last decade with great success.

These past years a lot of effort has been invested in integrating metabolic and trancriptional networks (Herrgard et al. 2006; Simeonidis, Chandrasekaran, and Price 2013).

Since the analysis carried out in this thesis used protein interaction and metabolic networks as objects of study, the next section will describe them in more detail.

Figure 2.6. Gene regulatory network associated to the glycolysis pathway in Escherichia coli K12 MG1655 from the EcoCyc database.