STTR Phase II: Artificial Intelligence (AI)-based Development of Neutralizing Antibodies for SARS-CoV-2

这项小企业创新研究（SBIR）项目的更广泛影响/商业潜力将导致开发针对SARS-CoV-2病毒的工程中和抗体，这些抗体可用作治疗剂，以减轻COVID-19感染的严重程度，减少住院和疾病进展的机会。随着SARS-CoV-2病毒继续变异，有必要通过产生广泛的中和抗体来加强集体防范，这些抗体可以单独覆盖一系列变体（Delta、Beta、Omicron）。这些抗体，当作为一种抗体鸡尾酒给药时，可能对不同人群的变异提供广泛的保护。这个项目的方法产生了中和抗体，这些抗体是经过特殊设计的，可以与刺突蛋白的不同区域结合，从而增加了一种或多种工程抗体有效对抗未来突变病毒的可能性。高通量筛选、下一代测序和基于人工智能（AI）的抗体设计相结合，可以系统地探索广泛的抗体序列。这种方法还具有工程抗体的好处，这些抗体更有效，更容易管理，在具有挑战性的环境条件下更稳定，制造成本更低，从而使治疗方法更容易分发给低收入国家。该STTR二期项目提出通过提供所需的抗体序列突变结合数据来实现人工智能和机器学习抗体工程方法。目前可用的抗体数据集有数千个数据点，该团队建议生成数千万个数据点的数据集。该项目还将生成阳性和阴性抗体结合数据，可能导致更高性能的学习抗体结合模型。该项目旨在测试一种假设，即合成抗体可以等于或优于自然产生的抗体，以中和SARS-CoV-2的传染性。这种方法可能会产生广泛的抗体变异。这种基于人工智能的抗体工程的应用将集中在通过酵母展示、高通量荧光活化细胞分选（FACS）和下一代测序相结合，发现针对SARS-CoV-2刺突蛋白多个不同区域的大量高亲和力中和抗体。将这些高通量数据生成工作流程与最新的深度神经网络相结合，可能会产生一种新的方法，可以有效地发现针对当前和未来大流行的高效抗体。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) project will lead to the development of engineered neutralizing antibodies for the SARS-CoV-2 virus that can be used as therapeutic agents to diminish the severity of a COVID-19 infection and decrease the chances of hospitalization and progressive disease. As the SARS-CoV-2 virus continues to mutate it is necessary to increase collective preparedness by generating a wide collection of neutralizing antibodies that individually provide coverage for a range of variants (Delta, Beta, Omicron). These antibodies, when administered as an antibody cocktail, may offer broad protection over a diverse population of variants. This project’s approach generates neutralizing antibodies that are specifically engineered to bind to different regions of the spike protein thereby increasing the probability that one or more of the engineered antibodies will be effective against future mutated versions of the virus. The proposed combination of high-throughput screening, next-generation-sequencing and artificial intelligence (AI)-based antibody design allows systematic exploration of vast ranges of antibody sequences. This approach also has the benefit of engineering antibodies that are more potent, easier to administer, more stable under challenging environmental conditions, and less costly to manufacture, leading to therapeutics that can be more readily distributed to low-income countries.This STTR Phase II project proposes to enable AI and machine learning antibody engineering approaches by providing needed antibody sequence mutation binding data. Currently available antibody datasets number in the thousands of datapoints and the team proposes to generate datasets that number in the tens of millions. The project will also be generating both positive and negative antibody binding data, potentially leading to higher performing learned antibody binding models. This project seeks to test the hypothesis that synthetic antibodies can be the equal of, or better than, naturally occurring antibodies for neutralizing SARS-CoV-2 infectivity. This approach could potentially develop a wide range of antibody variations. The application of this AI-based antibody engineering will be focused on discovering a large array of high-affinity neutralizing antibodies targeting multiple, different regions of the SARS-CoV-2 spike protein through the combination of yeast-display, high-throughput fluorescence-activated cell sorting (FACS) and next generation sequencing. Combining these high-throughput data generation workflows with the latest deep neural networks may lead to a new methodology that can efficiently discover high performing antibodies for the current pandemic and those in the future.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.