CCI Phase I: NSF Center for Computer Assisted Synthesis

NSF计算机辅助合成中心(C-CAS)是由化学部化学创新中心(CCI)项目支持的。这个一期中心由圣母大学的奥拉夫·威斯特领导。其他研究小组成员包括圣母大学的Nitesh Chawla、普林斯顿大学的Abigail Doyle、科罗拉多州立大学的Robert Paton、加州大学伯克利分校的Richmond Sarpong和犹他大学的Matthew Sigman。C-CAS的目标是将数据科学、机器学习、人工智能、化学反应和选择性优化、计算化学和有机合成相结合，以改变复杂有机分子合成的计划和执行方式。该中心将改变化学合成，提高美国境内制药、化学和技术行业的经济和社会竞争力。研究工作通过多个行业伙伴关系得到加强。因此，将培训新一代跨学科的“数据化学家”和机器学习学者团队，以应对现代合成化学的挑战和需求。将建立一个研究、网络和专业发展机会网络，以弥合学科之间的差距，吸引包括残疾学生在内的各种背景的学生。将利用社交媒体和大众媒体让公众参与进来。C-CAS的愿景是创建一个全面的平台，用于计算计划和优化合成路线，预测反应性能，并解决复杂分子合成固有的选择性(例如，化学、区域、立体)挑战。这个中锋有三个冲刺。第一个推动力是使用表示学习来统一来自各种来源的异类数据，包括来自文献、专利、高通量实验、电子实验室笔记本和高通量计算的无偏见、密集和“干净”的微观和宏观数据。第二个重点是利用这种统一的数据表示来解决化学反应优化的“箭头”问题，这是大多数合成中的速度限制步骤，通过将产生可解释和可解释的统计和机器学习模型的主动和转移学习来解决。第三个重点是展示异质数据和优化算法与现有合成规划程序的组合使用，以对可能的合成路线进行评分，并为合成化学家提供复杂分子合成中进行/不进行的决策。C-CAS为新一代“数据化学家”和机器学习研究人员提供量化培训，从而使他们能够参与跨学科科学，包括产学研合作。该中心提供在线和面对面研讨会、网络以及制定和分享这一快速发展领域的最佳实践等方案，以增强跨学科经验和社会影响。C-CAS的数据科学和计算部分提供了非传统的研究和联网机会，特别适合残疾学生和代表性不足的少数族裔学生。C-CAS通过大众和社交媒体以及面对面的交流，让非科学家参与到化学和机器学习的整合讨论中。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The NSF Center for Computer Assisted Synthesis (C-CAS) is supported by the Centers for Chemical Innovation (CCI) Program of the Division of Chemistry. This Phase I Center is led by Olaf Wiest of the University of Notre Dame. Other team members include Nitesh Chawla, also of the University of Notre Dame, Abigail Doyle of Princeton University, Robert Paton of Colorado State University, Richmond Sarpong of the University of California, Berkeley, and Matthew Sigman of the University of Utah. The goal of C-CAS is to combine data science, machine learning, artificial intelligence, chemical reactions and selectivity optimization, computational chemistry, and organic synthesis to transform how the synthesis of complex organic molecules is planned and executed. This center will transform chemical synthesis and increase the economic and societal competitiveness of pharmaceutical, chemical, and technology industries within the US. The research efforts are enhanced through multiple industrial partnerships. As a result, a new generation of interdisciplinary teams of "data chemists" and machine learning scholars will be trained to address the challenges and demands of modern synthetic chemistry. A network of research, networking and professional development opportunities will be established to bridge the gap between disciplines, drawing students from a broad range of backgrounds including students with disabilities. Social and mass media will be utilized to engage the general public. The vision of C-CAS is to create a comprehensive platform for computationally planning and optimizing synthetic pathways, predicting reaction performance, and addressing the selectivity (e.g., chemo-, regio, stereo-) challenges inherent in complex molecule synthesis. This center has three thrusts. The first thrust is to use representation learning to unify heterogeneous data from a variety of sources, including unbiased, dense, and "clean" microscopic and macroscopic data from the literature, patents, high-throughput experimentation, electronic laboratory notebooks and high-throughput computation. The second thrust is to exploit this unified data representation to address the "over-the arrow" problem of chemical reaction optimization that is the rate-limiting step in most syntheses, through active and transfer learning that will yield interpretable and explainable statistical and machine learning models. The third thrust is to demonstrate the combined use of heterogeneous data and optimization algorithms with existing synthesis planning programs to score possible synthetic routes and provide synthetic chemists with go/no-go decisions in the synthesis of a complex molecule. C-CAS provides quantitative training to a new generation of "data chemists" and to machine learning researchers, thereby equipping them for participation in interdisciplinary science, including industry-academic collaborations. The Center delivers programming such as online and in-person workshops, networking, and formulation and sharing of best practices in this rapidly evolving field that enhances the interdisciplinary experience and societal impact. The data science and computational portion in C-CAS provides non-traditional research and networking opportunities that are particularly suitable to students with disabilities and underrepresented minority students. C-CAS engages non-scientists in discussions of the integration of chemistry and machine learning through mass- and social media as well as in-person communications.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.