Machine-learning generated nucleases for accelerating the deployment of a novel, low-emission food production systems

**伊甸园生物公司提出的工具将通过利用最先进的机器学习技术来支持可持续的蛋白质生产，1)使食品企业能够扩大其活动规模，2)加速净零倡议。**精密发酵(PF)是一种不使用动物的可持续蛋白质生产方法。PF具有巨大的潜力，是重大投资的重点。这种方法通过在微生物中创建“细胞工厂”来创造功能成分。通过PF可以生产酶、调味剂、蛋白质、维生素、天然色素和脂肪，市场正在迅速增长(预计到2030年将达到373.5亿美元)。这一增长的关键驱动力之一是全球对可持续蛋白质生产的需求尚未得到满足。越来越多的公司生产依赖PF的蛋白质(例如，食品技术初创公司)，然而，它们都难以实现工业化规模的产品生产。这不仅给这些公司创造收入带来了巨大的障碍，还减缓了它们经济部门的现代化和向可持续发展的过渡。PF涉及通过基因修饰微生物生产蛋白质，因此，它严重依赖基因组编辑工具和核酸酶，核酸酶是一种在特定位置切割基因组以进行编辑的蛋白质。目前可用的核酸酶数据集要么限于非常特定的、适用范围很窄的类型，要么条目有限，而商业上可用的核酸酶的使用昂贵得令人望而却步。为了继续提供同类最佳的菌株优化服务，EB建议开发一个完全注释的核酸酶储存库和一个ML工具，用于生成可根据客户的特定需求定制的非IP侵权核酸酶。这一工具将为EB的服务增加重大价值，并将加速PF公司的规模和向可持续食品生产系统的过渡。

**Eden Bio's proposed tool will support sustainable protein production by leveraging state-of-the-art machine learning technology, 1) enabling food businesses to scale up their activities and 2) accelerating Net Zero initiatives.**Precision fermentation (PF) is a sustainable protein production method that doesn't use animals. PF has tremendous potential and is a focus of significant investments. This method creates functional ingredients by creating "cell factories" in microbes. Enzymes, flavouring agents, proteins, vitamins, natural pigments, and fats can be produced through PF, and the market is growing rapidly (predicted to reach $37.35B by 2030). One of the key drivers of this growth is the unmet global demand for sustainable protein production.A growing number of companies produce proteins (e.g., food-tech startups) that rely on PF, however, they all struggle to produce their products at an industrial scale. This not only presents huge barriers for these companies to reaching revenue generation but also slows down their economic sector's modernisation and transition towards sustainability.PF involves the production of proteins by genetically modifying microorganisms, thus, it is heavily reliant on genome editing tools and nucleases, a type of protein that cuts the genome in a specific place to allow an edit to take place. Currently available nuclease datasets are limited to either very specific, narrowly applicable types or have limited entries, while the use of commercially available nucleases is prohibitively expensive.To continue offering best-in-class strain optimisation services, EB proposes to develop a repository of fully annotated nucleases and an ML tool for generating non-IP infringing nucleases that will be customisable for the clients' specific needs. This tool will add significant value to EB's services and will allow for the acceleration of PF company's scaling and transition to sustainable food production systems.