CAREER: Advancing evolutionary genomics and eukaryotic biodiversity research through accurate, scalable, and flexible frameworks for structural genome annotation

高质量的注释和相关的基因组组装对于了解给定生物体中的基因如何工作是必要的。与基因及其结构相关的变异为研究形态、生理和行为性状提供了一个框架。 在高通量测序时代，所尝试的基因组的大小和复杂性急剧增加。尽管如此，超过91%的基因组包含大量的基因注释错误。 地球生物基因组计划打算在未来十年内测序150万个真核生物基因组。相关项目，如脊椎动物基因组计划，全球无脊椎动物基因组联盟和10,000植物基因组计划将有助于令人兴奋的基因组对生物多样性研究的贡献。可靠、高效和集成良好的软件，保持与社区数据标准的连接，对于处理这些举措产生的大量数据至关重要。 EASEL（Efficient，Accurate，Scalable Eukaryotic modeLs）框架将极大地减轻研究人员的负担，其中许多人正试图用小团队组装和注释基因组。 与这些小团队的合作将支持注释平台的开发，该平台在用户友好的包中实现机器学习。 同时，EASEL将通过响应更大和更复杂的基因组的需求来提高效率和准确性。 与这些大规模计划的合作将支持密集的本科生实习计划，将生物学学生与基因组注释背景下的大数据，生物信息学和机器学习联系起来。EASEL（Efficient，Accurate，Scalable Eukaryotic modeLs）是一个集成的、可访问的深度学习框架，用于在有限或广泛的外部证据的情况下注释真核生物参考基因组。 该软件将通过完整的工作流程改进基于证据和从头推导的基因模型，包括通过基因模型注释进行重复识别。 软件开发将与代表30多个新的真核生物基因组的研究伙伴关系配对，包括昆虫，植物和动物。在成功实施后，EASEL将转化为与Galaxy Toolshed兼容的框架，以便可以通过任何本地实例自由安装和执行。 将开发一个Tripal/Galaxy数据库模块，用于安装在任何Tripal进化枝或模式生物网上储存库，以提供接近社区数据库中基因组资源的分析能力。 数据库一级的整合将首先在林木基因组学和表型组学资源TreeGenes内进行评估。软件开发将被整合到多学科研究和教育驱动的模型中。一个新的本科生暑期培训机会，基因组组装和注释，将引导学生通过三个模块的研究经验，最终将在一个注释马拉松和总共十个新的基因组注释。 该项目的软件和成果将在这里分发：https://gitlab.com/PlantGenomicsLab/HBEF/-/tree/master/AnnotationThis奖项反映了NSF的法定使命，并被认为值得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来支持。

A high-quality annotation and associated genome assembly are necessary to understand how genes in a given organism work. Variation associated with genes, and their structure, provides a framework for examining morphological, physiological, and behavioral traits. In the era of high-throughput sequencing, the size and complexity of the genomes attempted has dramatically increased. Despite this, over 91% of these genomes contain a multitude of gene annotation errors. The Earth BioGenome Project intends to sequence 1.5 M Eukaryotic genomes in the next ten years. Related projects, such as the Vertebrate Genomes Project, the Global Invertebrate Genome Alliance, and the 10,000 Plant Genomes Project will contribute to exciting genomic contributions to biodiversity research. Reliable, efficient, and well-integrated software, that maintain connectivity to community data standards, will be critical to address the tremendous data generated by these initiatives. The EASEL (Efficient, Accurate, Scalable Eukaryotic modeLs) framework will tremendously ease the burden on researchers, many of whom, are attempting to assemble and annotate genomes with small teams. Collaborations with these small teams will support the development of an annotation platform that implements machine learning in a user-friendly package. At the same time, EASEL will improve the efficiency and accuracy by responding to the needs of larger and more complex genomes. Collaborations with these large-scale initiatives will support an intensive undergraduate internship program to connect biology students to big data, bioinformatics, and machine learning in the context of genome annotation. EASEL (Efficient, Accurate, Scalable Eukaryotic modeLs), an integrated and accessible deep learning framework for the annotation of eukaryotic reference genomes with limited or extensive external evidence, will be developed. The software will improve both evidence-based and ab initio derived gene models through a full workflow, that encompasses repeat identification through gene model annotation. Software development will be paired with research partnerships representing over 30 new eukaryotic genomes, including insects, plants, and animals. Following successful implementation, EASEL will be translated into a framework compatible with the Galaxy Toolshed so that it can be freely installed and executed through any local instance. A Tripal/Galaxy database module will be developed for installation on any Tripal clade or model organism web-based repository to provide analytical capacity in proximity to the genomic resources housed in community databases. Integration at the database level will be evaluated first within the forest tree genomics and phenomics resource, TreeGenes. Software development will be integrated in a multi-disciplinary research and education driven model. A new undergraduate summer training opportunity, Genome Assembly and Annotation, will guide students through a three module research experience that will culminate in an Annotation-thon and a total of ten new genome annotations. Software and results from this project will be distributed here: https://gitlab.com/PlantGenomicsLab/HBEF/-/tree/master/AnnotationThis 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.