Currently, most design in synthetic biology is done manually by a domain expert on an ad-hoc basis. This practice, however, hinders large-scale design. One of the major challenges in biological systems design is that molecular interactions on the genome scale are extremely complicated. Synthetic biologists can no longer rely on manual effort for designing synthetic systems at a genome-wide scale. As the complexity of systems design increases, it is not an option but a necessity to utilise computer aided or automated design processes.
Design automation starts from programmatic analyses and therefore requires a model space representation of the design domain. Much effort has already been put into devising mathematical models of genetic circuits in order to represent the inner works of biological systems and to predict their behaviour in silico. Many aspects of molecular biology can be incorporated into mathematical models. However, it is often difficult to obtain appropriate parameter values relevant to the context specific aspects of biology, such as the expression of specific genes at transcriptional and translational levels, that are important to the accuracy of models in varying environmental conditions. There is a limited availability of well characterized genetic parts for use in mathematical models. Consequently, models either only account for limited but known conditions or have to rely on arbitrary parameter values for the unknowns. It is the choice of making models either accurately limited or inaccurately sophisticated. Therefore, the in vivo implementations of model-based designs nearly always end up not exhibiting the intended or predicted outcome.
Directed evolution can be employed as a rescue for the inherent inaccuracy in models. This project aims to develop an effective and systematic design automation process by employing directed evolution. Random mutagenesis and selective pressure are applied to microbes in a controlled manner with a goal to engineer a desired phenotype. Computer models, based on partial information, can be used to initially design corresponding genetic constructs. Directed evolution will then find solutions, refinement of design and model compositions, based on reasoning over the selection of appropriate in vivo measurement methods, sampling points and their analyses in relation to the in silico counterpart.