LOOPER: Machine learning for druggable protein conformations

该项目旨在结合分子动力学(MD)模拟和机器学习(ML)技术来预测蛋白质的运动方式，并改进类药物分子针对灵活蛋白质区域的方式。这是建立在Degiacomi小组在生成神经网络(GNN)和MEY小组在MD模拟和药物设计方法开发方面的专业知识的基础上的。学生将回答以下问题：GNN能否用于产生生物相关的蛋白质构象？那么，如何利用这些结构来改进针对IDR的类药物分子的设计呢？这些问题是理解帕金森氏症和阿尔茨海默氏症等神经退行性疾病以及抗菌素耐药性背后的分子机制的核心。为期三个月的ReDesign Science实习课程(www.reDesigncience.com)将使学生能够应用所学的MD和ML技术来洞察一个活跃的药物发现项目，并学习如何根据行业时间表简化他们的工作。

This project aims at combining molecular dynamics (MD) simulation and machine learning (ML) techniques to predict how proteins move, and to improve the way drug-like molecules can target flexible protein regions. This builds upon the expertise of the Degiacomi group on generative neural networks (GNNs) and that of the Mey group in method development for MD simulations and drug design. The student will address the questions of: can a GNN be used for generating biologically relevant protein conformations? How can the ensemble of structures then be used for improved design of drug-like molecules that specifically target their IDRs? These questions are central to understanding neurodegenerative diseases such as Parkison's and Alzheimer's, as well as the molecular mechanisms behind antimicrobial resistance. A three-month placement at Redesign Science (www.redesignscience.com) will enable the student to apply their acquired MD and ML techniques to get insights into an active drug discovery project, and learn how to streamline their work in accordance with industry timescales.