The genesis of the bottom-up approach to systems biology was the availability of the first full genome sequences. In principle, these sequences had information about all the genetic elements that underlie the function of the sequenced organism. Enough information was available about the function of subsets of these genes – namely the genes encoding metabolic functions – that an organized assembly of all the biochemical, genetic, and genomic information was achievable. Such an organized assembly is de facto a knowledge base, or a k-base, that gives rise to a network reconstruction at the genome-scale. Since such reconstructions are represented with accurate chemical equations, they can be mathematically described. A mathematical description can be used to compute functional states of a network that correspond to observable phenotypes and biological functions. With these elements in place, a new genome-scale science was born that focused on mechanistic genotype–phenotype relationships.

The first genome-scale models of metabolism appeared in 1999 and 2000. In the next half-decade or so, an enthusiastic group of investigators developed many fundamental concepts, in silico methods, and algorithms to analyze their properties. At times, and to many, these initial efforts seemed mostly exploratory. Fortunately, in the mid-2000s an abundance of data sets and data types became available to validate and demonstrate the utility of genome-scale models for research and discovery. At the end of the decade several highly curated models were available for model organisms. These models gained predictive power and over the next 5 years or so, a number of prospective uses of genome-scale models appeared. In other words, predictive genotype–phenotype relationships had appeared. These predictions were somewhat limited in scope, but proved useful for a series of applications. Currently the range of possible predictions of biological properties and functions is growing rapidly and it appears that this approach to genome-scale science is in its early stages of development, with a bright future ahead of it. For most of this fifteen-year history, the focus of genome-scale models has been metabolism. After initial successes with metabolic genome-scale models, it became clear that the same approach that led to their genesis could be applied to any other cellular process reconstructed in biochemically accurate detail.