Computational design of novel proteins to bind unnatural cofactors

广告项目描述（最多500字）。这将是向未来学生宣传的文本：从头蛋白质设计正在迅速成为创造新蛋白质结构的可行策略，尽管为这些结构添加功能仍然具有挑战性[1]。虽然在设计新型酶方面取得了一些成功，但它们的酶活性福尔斯达不到大多数天然系统的水平，因此有必要进行定向进化以提高活性[2]。在自然界中，辅因子通常被掺入蛋白质中以增加大量的功能，因此它们提供了一种有吸引力的途径来创造高活性的酶。有超过70，000种实验确定的酶结构，其中约12，000种含有辅因子[3]。该数据集包含大量信息，可用于设计结合新型辅因子的蛋白质。该项目的目的是应用Wood实验室正在开发的基于深度学习的工具，设计可以结合小分子催化剂作为非天然辅因子的新型蛋白质。我们靶向的分子是高度活性的生物相容性催化剂，通过将它们与蛋白质结合，我们的目标是调节底物特异性以及产物的对映体选择性。设计策略部署了深度学习来识别小分子中与天然辅因子化学相似的区域，并利用结合辅因子的蛋白质的已知结构信息来指导设计过程。大量的设计将通过计算生成和筛选，然后利用爱丁堡基因组铸造厂提供的机器人设备，将最有前途的候选人进行大规模的实验验证。在进行详细的催化、生物物理和结构分析之前，将筛选设计的辅因子结合。在这个过程中产生的信息将被纳入后续的设计，创造一个工程“设计，建造，测试”周期。在整个项目中，学生将接受实验自动化，分子生物学，生物物理技术和结构表征的培训。此外，华莱士实验室将支持设计酶的生化表征培训。虽然这个项目的主要重点是实验性的，但学生将有足够的机会发展机器学习和蛋白质原子模拟的计算技能。不需要计算技术的先验知识，因为学生将在发展这些技能方面得到很好的支持，但使用Python编程语言进行基本编程的经验将是有利的。

Project Description for Advert (max 500 words). This will be the text that is advertised to prospective students:De novo protein design is quickly becoming a viable strategy for creating novel protein structures, although adding functionality to these structures remains challenging [1]. While there has been some success in designing novel enzymes, their enzymatic activity falls short of most natural systems, making it necessary to perform directed evolution to improve activity [2]. In nature, cofactors are often incorporated into proteins to add a vast array of functionality, and so they offer an attractive route to creating highly active enzymes. There are > 70,000 experimentally determined structures of enzymes, and of these around 12,000 contain cofactors [3]. This dataset contains a huge amount of information that could be utilised to design proteins that bind novel cofactors.The aim of this project is to apply deep learning-based tools, which are being developed in the Wood lab, to design novel proteins that can bind small-molecule catalysts as unnatural cofactors. The molecules that we are targeting are highly active, biocompatible catalysts, and by binding them to a protein, we aim to tune substrate specificity, as well as enantioselectivity of the products. The design strategy deploys deep learning to identify regions of the small molecule that are chemically similar to natural cofactors, and exploits known structural information for proteins that bind cofactors to guide the design process. Large numbers of designs will be generated and screened computationally, before the most promising candidates are taken forward for experimental validation at scale, taking advantage of the robotics facilities available through the Edinburgh Genome Foundry. Designs will be screened for cofactor binding before a subset are taken forward for detailed catalytic, biophysical and structural analysis. Information generated during this process will be incorporated into subsequent rounds of design, creating an engineering "design, build, test" cycle.Throughout this project, the student will receive training in experimental automation, molecular biology, biophysical techniques, and structural characterisation. Furthermore, the Wallace lab will support training in biochemical characterisation of the designed enzymes. While the major focus of this project is experimental, there will be ample opportunity for the student to develop computational skills in machine learning and atomistic simulation of proteins. No prior knowledge of computational techniques is required, as the student will be well supported in developing these skills, but experience of basic programming with the Python programming language would be advantageous.