Antonia Mey, antonia.mey@ed.ac.uk
Andrea Weisse, Andrea.Weisse@ed.ac.uk
Project Description
As part of ongoing bacterial and fungal warfare in nature, bacteria have evolved to express proteins to destroy antibacterial agents secreted by fungi, such as penicillin. Humans have long exploited these antibacterial agents as therapeutics in the form of antibiotics, drastically reducing mortality from infectious diseases. Overuse of antibiotics has promoted the emergence of resistance in bacteria, leading to the current antimicrobial resistance (AMR) crisis [1].
One way in which bacteria acquire resistance is through the expression of ?-lactamases, an enzyme that attacks a common pattern in antibiotics. We are particularly interested in a class of ?-lactamases called metallo ?-lactamases (MBL). MBLs belong to a large group of proteins (~30% of all proteins) called metalloproteins that require metal ions to carry out their biological function. Aside from providing bacteria with antibiotic resistance, metalloproteins provide numerous important biological functions, such as signal transduction and energy storage.
In this project, we will develop a machine learning model for metalloproteins that will provide quantum mechanical accuracy to model enzyme inhibition and mutational effects, replacing costly quantum mechanical calculations with machine learning models. We will be leveraging data generated through QCArchive and the OpenForceField initiative and build on our own conditioned diffusion models SILVR [2] to identify important mutational sites and guide the design of new ways to inhibit MBLs.
