TABLE 1.

Characteristics of various network types discussed in this review and their main advantages and disadvantages.

Network type	Node	Edge	Required information to build the network	Pros	Cons
Protein-protein interaction networks	Proteins	Physical interactions	Measurement of the actual protein interactions e.g. using yeast two-hybrid, affinity purification mass spectrometry or small-scale binding experiments	Many different resources, based on physical interactions ensuring larger coverage	Highly incomplete, biases in network generating methods
Metabolic networks	Metabolites	Enzymes, reactions	Measured reactions of the enzymes	Most complete network type, good for systematic modelling	Need to decide what parameter to optimise
Gene regulatory networks	Transcription factors, promoters, enhancers, and target genes	Regulatory interaction	Measurement or modelling of the regulatory interactions e.g. using ChIP-seq, yeast one-hybrid, or through inference from transcriptomics	Various network building approaches to build large coverage and make it research question specific	Highly variable and state-specific, cannot infer feedback loops from transcriptomics only
Gene co-expression networks	Genes	Similarity between the expression of two genes	Gene expression measurement	Needs only transcriptomic data	Correlation does not always equal causation