Combining Machine Learning, Molecular Dynamics and Membrane Biophysics to identify new therapeutics for the treatment of Tuberculosis

结核病目前是世界上主要的死亡原因之一，仅2017年就报告了1000万新病例，130万例死亡（世卫组织2018年全球结核病报告），另一个复杂因素是多药和完全耐药菌株的演变。迫切需要开发有效的新型结核病治疗剂，而在这一过程中关键是确定合适的靶向蛋白。MmpL3是一种跨膜蛋白，对细菌细胞的复制和生存能力至关重要，因此代表了一个合适的靶标。最近对耻垢分枝杆菌MmpL3结构的测定（Cell 2019; 176: 636-648）为开发新的治疗策略提供了起点。分子动力学（MD）模拟将用于构建结核分枝杆菌（Mtb）的MmpL3模型，促进药物-蛋白质相互作用的研究，已知的抑制剂将在生理条件下建模，蛋白质嵌入细胞膜的现实表现。计算模型的验证将通过x射线衍射和光学显微镜对Mtb MmpL3嵌入的模型膜的结构和力学进行研究来实现。确定已知抑制剂与Mtb MmpL3和已知耐药突变体的结合模式将被用作机器学习（ML）的输入，以生成搜索大型化合物库的规则，特别是锌数据库，以确定合适的化合物进行筛选。该项目将为学生提供广泛的技能，计算建模，机器学习，蛋白质表达和纯化以及实验膜生物物理学。

Tuberculosis (TB) is currently one of the world's leading causes of mortality with 10 million new cases reported in 2017 alone and 1.3 million deaths (Global Tuberculosis Report 2018 WHO), a further complicating factor is the evolution of multi-drug and totally drug resistant strains. There is an urgent need to develop effective new therapeutic agents to target TB and critical in this process is the identification of a suitable protein to target. MmpL3 is a transmembrane protein which is essential for the replication and viability of bacterial cells and therefore represents a suitable target. The recent determination of the structure of MmpL3 from M. smegmatis (Cell 2019; 176: 636-648) provides the starting point for developing new therapeutic strategies. Molecular Dynamics (MD) simulations will be utilised to construct a model of MmpL3 for M. tuberculosis (Mtb) facilitating investigation of drug-protein interactions, known inhibitors will be modelled at physiological conditions with the protein embedded in a realistic representation of the cell membrane. Validation of the computational model will be achieved through the investigation of the structure and mechanics of model membranes, in which Mtb MmpL3 is embedded, via X-ray diffraction and light microscopy. Identification of the binding modes of know inhibitors to Mtb MmpL3 and known drug resistant mutants will be used as input into Machine Learning (ML) to generate rules to search large compound libraries, in particular the Zinc database, to identify suitable compounds to screen. This project will provide the student with a broad range of skills, computational modelling, machine learning, protein expression and purification and experimental membrane biophysics.