Development of machine learning methods for automated design of new biological functions in bacterial proteins.

开发机器学习方法，用于自动设计细菌蛋白质的新生物功能。

• 批准号: 2600923
• 金额: --
• 依托单位: Brunel University London
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2021
• 资助国家: 英国
• 起止时间: 2021 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2600923
• 关键词: Development; machine; learning; methods; automated

Protein design has matured to an exciting and leading area in synthetic biology, as confirmed by the 2018 Chemistry Nobel Prize to Frances Arnold. However, significant challenges remains in developing effective methods to automate design choices. In particular, there is no method to perform rational design of protein function by modification of protein internal dynamics. This research project is aimed at combining machine learning techniques and biomolecular simulations to develop the first AI-based 'protein engineer'. The doctoral researcher will work to: 1) design and implement a machine learning predictor of the effect of mutations on protein dynamics; 2) integrate the predictor into a widely used biomolecular simulation engine to automate selection and testing of candidate mutations; 3) incorporate the integrated machine learning simulation engine into a software container for easy deployment on different architectures. The optimal candidate should have a BSc and/or MSc degree in computer science, computational biology, computational chemistry or bioinformatics. Background in software development in Python and Linux/UNIX is expected.

蛋白质设计已经成熟为合成生物学中令人兴奋和领先的领域，正如2018年化学诺贝尔奖授予弗朗西斯·阿诺德所证实的那样。然而，在开发自动化设计选择的有效方法方面仍然存在重大挑战。特别地，没有通过修改蛋白质内部动力学来进行蛋白质功能的合理设计的方法。该研究项目旨在将机器学习技术和生物分子模拟相结合，开发第一个基于人工智能的“蛋白质工程师”。博士研究员将致力于：1）设计和实现突变对蛋白质动力学影响的机器学习预测器; 2）将预测器集成到广泛使用的生物分子模拟引擎中，以自动选择和测试候选突变; 3）将集成的机器学习模拟引擎集成到软件容器中，以便在不同架构上轻松部署。最佳候选人应具有计算机科学，计算生物学，计算化学或生物信息学的理学学士和/或理学硕士学位。具有Python和Linux/UNIX软件开发背景。