Types of biological networks


Biological networks

Different types of information can be represented in the shape of networks in order to model the cell (Figure 10). The meaning of the nodes and edges used in a network representation depends on the type of data used to build the network and this should be taken into account when analysing it.

Types of biological interactions that can be represented by networks

Figure 10 Types of biological interactions that can be represented by networks (Adapted with permission from slides by Dr. Nataša Pržulj, Imperial College, London).

Different types of data will also produce different general network characteristics in terms of connectivity, complexity and structure, where edges and nodes potentially convey multiple layers of information.

Some of the most common types of biological networks are:

1. Protein-protein interaction networks

2. Metabolic networks

3. Genetic interaction networks

4. Gene / transcriptional regulatory networks

5. Cell signalling networks

Protein-protein interaction networks.

A protein-protein interaction network.
Figure 11 A protein-protein interaction network.
  •   Represent the physical relationships between proteins. They are central to practically every process that takes place in the cell.
  •   Proteins are represented as nodes that are linked by undirected edges.
  •   We will look at protein-protein interaction networks in more detail later in the course.

Metabolic networks

Metabolic network.
Figure 12 A metabolic network.
  • Represent the biochemical reactions that allow an organism to grow, reproduce, respond to the environment and maintain its structure.
  • Metabolites and enzymes take the role of nodes and the reactions describing their transformations are represented as directed edges.
  • Edges can represent the direction of the metabolic flow or regulatory effects of a specific reaction.

Genetic interaction networks

Genetic interaction network
Figure 13 A genetic interaction network.
  • Genetic interaction is the synergistic phenomenon where the phenotype resulting from simultaneous mutations in two or more genes is significantly different from the phenotype that would result from adding the effects of the individual mutations.
  • Represent a functional relationship between different genes, rather than a physical one.
  • Genes are represented as nodes and their relationships as edges. Depending on the type of evidence behind the interaction, directionality can be inferred for the edges.

Gene / transcriptional regulatory networks

Gene / transcriptional regulation network
Figure 14 A gene / transcriptional regulatory networks.
  • Represent how gene expression is controlled.
  • Genes and transcription factors are represented as nodes, while the relationship between them is depicted by different types of directional edges.
  • Regulatory RNAs and other mechanisms can also form part of this type of network. 
  • Usually generated via databases representing consensus knowledge of gene regulation (e.g. Reactome or KEGG), although large-scale experimental datasets are increasingly available.

Cell signalling networks

Cell signalling network
Figure 15 A cell signalling network.
  • Cell signalling is the communication system that controls cellular activities.
  • Signalling pathways represent the ordered sequences of events and model the information flow within the cell.
  • Gene regulation networks can be considered as a sub-type of cell signalling networks, focusing on a specific signalling event which is often the final stage of a signalling cascade.
  • Elements in the pathway (e.g. proteins, nucleic acids, metabolites) are represented as nodes and the flow of information is represented by directed edges.
  • Are systematically represented by two types of resources:
    • Pathway databases (also known as 'process description' resources) such as Reactome, KEGG or Wikipathways. These aim to provide a formal representation of the current scientific consensus on cell signalling pathways. They are generated by manual curation and organise the information in the form of reactions, with substrates and products being affected by the action of catalysers. This information must be converted according to specific rules in order to be represented as a network. Some information loss can occur during this process.
    • Reaction network databases (also known as 'activity flow' resources) such as Signor, SignaLink or SPIKE. These aim to capture known binary relationships in cell signalling, such as activation, phosphorylation, etc. They are generally manually curated, but not always. In contrast with the pathway databases, they are already graphs in the mathematical sense and require no transformation in order to be represented as a network.