PIPP Phase I: Develop and Evaluate Computational Frameworks to Predict and Prevent Future Coronavirus Pandemics

一旦检测到新型冠状病毒或新变种，新型冠状病毒或新变种在人与人之间传播的可能性有多大？患者的病情会有多严重？什么样的新型冠状病毒可能导致未来的大流行？了解这些问题的答案可以帮助各国做出正确的战略决策。难题在于如何尽早预测新病毒的行为和致病严重程度。一组研究人员认为，他们已经找到了回答这些问题的方法，他们开发了新的人工智能软件工具，可以根据病毒的基因组序列预测病毒的行为。这组研究人员认识到机器学习应用中的潜在偏见，以及增加未来人工智能劳动力多样性的必要性。利用他们在基因组学、数据科学、人工智能、遗传学、传染病、化学工程、公共卫生和通信方面的专业知识，该研究团队将组织培训研讨会和活动，提供人工智能的文化响应教学，为教师提供数据科学培训，并为高中生提供与上下文相关的编码体验。该团队将通过社区外展促进公众对科学的信任和对错误信息的识别。该研究小组将基于生物学知识和假设建立一个深度学习模型原型，该模型可以从基因组序列中预测病毒致病适应性，以测试病毒致病性的潜在规则。该团队将探索几种方法来纠正病毒基因组监测中的采样偏差，以便准确估计病毒株的适应度。该团队将调查东南亚所有可用的蝙蝠冠状病毒基因组的突变和重组谱，并建立一个原型地理空间模型，以预测所有可用的蝙蝠冠状病毒的重组概率。利用他们在遗传学和大分子结构建模方面的专业知识，该团队将测试SARS-CoV-2中的几个候选基因，以寻找潜在的致病规律。根据这些试点项目的结果，该团队将能够估计新出现的SARS-CoV-2变体或另一种新型冠状病毒的致病性适合度。该奖项由流行病预防的跨部门预测情报第一阶段（PIPP）计划提供支持，该计划由生物科学（BIO）、计算机信息科学与工程（CISE）、工程（ENG）和社会、行为和经济科学（SBE）委员会共同资助。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。

Once a novel coronavirus or a new variant is detected, how likely would the novel coronavirus or new variant transmit from person to person, and how sick could patients become? What kind of new coronaviruses could cause future pandemics? Knowing the answers to these questions can help nations make proper strategic decisions. The dilemma is how to predict the behavior and pathogenic severity of new viruses as early as possible. A team of researchers thinks they have found ways to answer these questions by developing new artificial intelligence software tools to predict the virus’s behaviors based on its genome sequence. This team of researchers recognizes the potential bias in machine learning applications and the need to increase diversity in the future artificial intelligence workforce. Leveraging their expertise in genomics, data science, artificial intelligence, genetics, infectious disease, chemical engineering, public health, and communication, this team of researchers will organize training workshops and activities providing culturally responsive teaching of artificial intelligence, data science training to teachers, and context-relevant coding experiences to high school students. The team will promote public trust in science and discernment of misinformation through community outreach. This research team will prototype a deep learning model based on biological knowledge and hypotheses that can predict viral pathogenic fitness from genomic sequences to test the potential rules for viral pathogenicity. The team will explore several methods to correct the sampling bias in viral genomic surveillance in order to accurately estimate the fitness of a viral strain. The team will investigate the mutation and recombination profiles in all available bat coronavirus genomes from the Southeastern Asia and build a prototype geospatial model to predict the recombination probability for all available bat coronaviruses. Leveraging their expertise in genetics and macromolecular structure modeling, the team will test a few candidate genes in SARS-CoV-2 for potential pathogenic rules. Based on the outcomes of these pilot projects, the team will be able to estimate the pathogenic fitness of an emerging SARS-CoV-2 variant or another novel coronavirus. This award is supported by the cross-directorate Predictive Intelligence for Pandemic Prevention Phase I (PIPP) program, which is jointly funded by the Directorates for Biological Sciences (BIO), Computer Information Science and Engineering (CISE), Engineering (ENG) and Social, Behavioral and Economic Sciences (SBE).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.