ABSTRACT
Bioprocess industries aim to employ bacteria for the large-scale production of a particular compound. The bacteria, however, simply grow in numbers to maintain their population, and they need to be manipulated when used as factories to produce an industrially important product in excess of what they require for their growth. One of the techniques often used to manipulate and optimize the bacterial cell is metabolic engineering, which involves rewiring the metabolic network of the cell, so that it directs all of its resources towards the formation of that particular product. Imperative to this process is the identification of mutable targets in the genome of the cell. Machine learning techniques have numerous potential applications in the in silico manipulation of the cells, most notably by allowing us to predict the consequences of a particular change in the cell's genome or metabolism, either from an existing set of experimental data, or by evolving the metabolism in a stepwise manner. This chapter discusses the potential of machine learning in engineering the metabolism of the cell. It also discusses the software packages and algorithms currently developed for handling and mining the huge datasets produced from metabolomics experiments, while also using them for the purpose of metabolic engineering.
