NSF Postdoctoral Fellowship in Biology FY 2020: Building quantitative models for plant transcription using convolutional neural networks for de novo promoter design

本行动资助NSF国家植物基因组计划2020财年生物学博士后研究奖学金。该奖学金支持奖学金获得者在主办实验室的研究和培训计划，该奖学金获得者还提出了扩大生物学参与的计划。Andy Zhou博士的研究和培训计划的标题是“使用卷积神经网络为作物植物工程的从头启动子设计构建植物转录的定量模型”。该奖学金的主办机构是联合生物能源研究所、联合基因组研究所和加州大学戴维斯分校。赞助科学家是Patrick M. Shih博士和Ronan C. O 'Malley博士。通过DNA和RNA测序揭示植物的内部工作可以为如何更好地设计作物植物提供有价值的见解，并且对于满足社会对可再生能源和环境可持续性日益增长的需求至关重要。然而，这些测序数据集提供了无数的可分析模式，如果没有复杂的计算分析，通常很难破译。该项目将开发植物生物学的机器学习模型，并将这些模型直接应用于农业生物技术的开发。这些技术将有助于提高农作物的经济价值，并通过利用植物创新食品、燃料或化学品生产，使社会受益。该项目通过创建一个互动网站来探索项目成果，强调对研究员进行科学交流的培训。该研究员将担任导师的角色，以吸引来自不同背景和代表性不足的当地高中生和大学生对科学事业的兴趣。该项目旨在开发最先进的机器学习模型，用于模型和作物植物系统中的转录起始和染色质可及性，以重新定义植物基因表达的核心机制，并规避工程植物系统面临的挑战。该项目利用转座酶可及染色质测定（ATAC）和基因表达帽化分析（CAGE）测序数据来训练深度卷积神经网络，以预测拟南芥和番茄基因表达的重要特征。这些机器学习模型将用于寻找开放的染色质，以促进转基因作物的转化，可调的转基因表达启动子，并指导模型和作物植物系统中合成生物学的从头DNA部分的设计。生成的所有测序数据将托管在NCBI等公共可用的存储库上。分析代码和机器学习模型将上传到GitHub和Kipoi基因组学模型动物园。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。

This action funds an NSF National Plant Genome Initiative Postdoctoral Research Fellowship in Biology for FY 2020. The fellowship supports a research and training plan in a host laboratory for the Fellow who also presents a plan to broaden participation in biology. The title of the research and training plan for this fellowship to Dr. Andy Zhou is "Building quantitative models for plant transcription using convolutional neural networks for de novo promoter design for crop plant engineering". The host institutions for the fellowship are the Joint Bioenergy Institute, the Joint Genome Institute, and the University of California, Davis. The sponsoring scientists are Dr. Patrick M. Shih and Dr. Ronan C. O’Malley.Uncovering the inner workings of plants through DNA and RNA sequencing can provide valuable insight on how to better engineer crop plants and is essential in meeting the increasing societal demand for renewable energy and environmental sustainability. However, these sequencing datasets provide a myriad of analyzable patterns and are often difficult to decipher without sophisticated computational analysis. The project will develop machine-learning models for plant biology and apply these models directly to develop agricultural biotechnology. These technologies will assist in enhancing the economic value of crop plants and benefitting society through innovating food, fuel, or chemical production using plants. The project emphasizes training of the fellow in scientific communication of findings to the general public by creating an interactive website to explore the project findings. The fellow will engage in mentorship roles to garner interest in scientific careers for local high school and college students of diverse and underrepresented backgrounds. The project aims to develop state-of-the-art machine learning models for transcription initiation and chromatin accessibility in both model and crop plant systems in order to redefine core mechanisms of plant gene expression and circumvent the challenges facing engineering plant systems. The project leverages Assay for Transposase Accessible Chromatin (ATAC) and Capped Analysis of Gene Expression (CAGE) sequencing data to train deep convolutional neural networks to predict features important to gene expression in Arabidopsis and tomato. These machine learning models will be utilized for finding open chromatin to facilitate transgenic crop transformation, tunable promoters for transgene expression, and guide the design of de novo DNA parts for synthetic biology in model and crop plant systems. All sequencing data generated will be hosted on publicly-available repositories such as NCBI. Analysis code and machine learning models will be uploaded to GitHub and Kipoi model zoo for genomics.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.