In this project, we aim to computationally optimise inhibitors for the enzyme DXS from the MEP pathway. This enzyme is essential for the biosynthesis of isoprenoid precursors for pathogens such as Mycobacterium tuberculosis; however, it does not exist in humans. To achieve our goal, we develop Deep Learning models trained in the abundance of publicly available chemical data. Those models can capture meaningful associations and gain knowledge of the chemical space. The knowledge can be transferred and applied to supervised, drug-related tasks enabling us to predict more accurately molecular properties like the physicochemical ones. Moreover, our objective is to deploy not only efficient but also interpretable models that allow getting a better understanding of the efficacy of the drugs. Interpretability is the key to determining the structural characteristics and the chemical properties that make molecules successfully inhibit DXS enzyme. Consequently, we aim to gain valuable information and develop tools to facilitate drug development. The extracted information and our predictions will be provided to our collaborators to synthesise and experimentally characterise a series of compounds. We will utilise their feedback to enhance further the efficiency and the usefulness of our work in this field.
Name of Doctoral Researcher
Georgios Kallergis
Name of Supervisor
Prof Dr Alice McHardy
Institute / Department
Department of Computational Biology Infection Research, HZI
Contact details
georgios.kallergis@helmholtz-hzi.de