Metabolism is one of the most dynamic and adaptive cellular processes. Much of this activity can now be assessed using metabolomics, which studies metabolism at the genome-scale. Despite providing one of the most direct readouts of phenotypic state available, to date there has been surprisingly little application of metabolomics to metabolic engineering strain design. Similarly, many of the mathematical and computational methods that have been widely used for strain design in metabolic engineering, such as constraint-based modeling strategies, also make little use of metabolomics data. In order to address this gap, we are developing new mathematical and computational strategies that use metabolomics datasets to improve the accuracy of strains designed in silico. We will first construct constraint-based models that predict intracellular metabolite concentrations in addition to the intracellular fluxes that are typically calculated. Combined with machine learning methods and metabolomics datasets, regulatory interactions can be inferred and included as new constraints in the model as well. The model will then be used to design a strain of S. cerevisiae, and the design implemented to assess the model’s predictive accuracy and to allow for iterative refinement.
Related Publications:
- LK-DFBA: a linear programming-based modeling strategy for capturing dynamics and metabolite-dependent regulation in metabolism
- NS-kNN: a modified k-nearest neighbors approach for imputing metabolomics data
- Improved metabolite profile smoothing for flux estimation
- Systematic Applications of Metabolomics in Metabolic Engineering