The usage of Genome-scale Metabolic models (GEMs) spans various applications across diverse fields. These models depict a metabolic network that represents the complete metabolism of a specific organism. The automated reconstruction of these models can be facilitated using tools like merlin. However, the draft models generated often contain gaps, primarily due to knowledge limitations in the databases that enable the reconstruction of these models (e.g., KEGG, ModelSEED), and due to wrong and faulty genome annotations. Certain tools, such as BioISO, aim at identifying metabolites that cannot be produced by a metabolic network. Other tools like Meneco try to discern a set of reactions that can be integrated into the model, aiming to rectify the existing gaps in the metabolic network. The execution time for these tools increases proportionally with the complexity of the metabolic network under scrutiny, potentially leading to prolonged execution times to address the possible gaps present in the model. Thus, there arises a need to develop a workflow that is both efficient and offers reliable solutions. In this study, a workflow integrating BioISO and Meneco (BioMeneco) was developed, coupled with the development of pertinent methods, with the aim of automating the process as much as possible, reducing the search space, and optimising the gap-filling process. The outcomes of the developed workflow were promising. It not only offered reduced execution times but also provided the capability for better refinement for various models when compared to a typical Meneco workflow. Regardless, while the developed workflow demonstrated efficiency, it also highlighted the challenges of relying on a single database and the complexities of metabolic networks, paving for further improvements and research in this domain.

Metabolic Transformation Algorithm (MTA) is a computational method that identifies potential metabolic interventions to shift a biological system from an undesired state to a desired one by targeting specific genes or reactions within a metabolic network. The Robust Metabolic Transformation Algorithm (rMTA) enhances this capability by incorporating worst-case scenario analyses and integrating the Minimization Of Metabolic Adjustment (MOMA) approach. By transitioning these algorithms from MATLAB to Python, this work aims to enhance their accessibility and usability within the scientific community. The Python implementations were validated by successfully replicating results from the original MATLAB implementation, including gene knockouts of RRM1 and RRM2. Furthermore, the rMTA algorithm was applied to an aging-related scenario in Caenorhabditis elegans to identify candidate genes and pathways involved in the metabolic response to unc-62 knockdown. Additionally, this work integrates an Evolutionary Algorithms (EAs)-based approach to rMTA to evaluate the effects of multiple gene knockouts simultaneously. This extension optimizes the identification of metabolic targets for phenotype reversion, demonstrating the potential for more effective therapeutic strategies. In summary, this dissertation enhances the accessibility of MTA and rMTA by implementing them in Python, applying them to an aging-related scenario, and extending their capabilities through the integration of EAs.