GRK 2403: Dissecting and Reengineering the Regulatory Genome

使生物体能够从一个基因组分化出多种细胞类型的复杂发育过程是由基因调控的复杂性编码的，而不是大量基因的可用性。人类基因组的完成使我们能够破译控制基因调控的过程，从三维染色体折叠的背景到特定转录因子的作用，最后到其产物RNA或蛋白质。解开多尺度遗传调控密码需要实验方法，越来越能够查询调控功能在全基因组和高通量水平。从单个DNA位点的体外效应到整个调控区的体内影响，现在可以从单个细胞中获得广泛的读数。基因组编辑可以改变基因组序列，并使用“表观遗传”修饰来扰乱调控区域的活性，由此产生的数据需要复杂的计算方法来解释结果并深入了解细胞调控过程。机器学习方法在从大型复杂数据中自动推断更高级别表示的能力方面取得了巨大的飞跃，例如在深度学习的背景下。结合实验和计算的方法将提供一个前所未有的机会，教下一代的研究人员在生物系统的背景下，基因组功能和基因调控的定量理解。由洪堡大学和杜克大学牵头的柏林机构之间拟议的联盟将通过由三个互补领域组成的国际研究培训小组（IRTG）来应对这一挑战。(1)高通量基因组学和编辑，（2）生物信息学和机器学习，以及（3）发育系统生物学。博士研究人员将在合作机构获得大量培训，从而受益于两个地点的协同作用和专业知识。在整个培训过程中，他们将得到来自大西洋两岸的计算和生物专家的共同建议。一个定制的计划将支持学生在弥合三个重点领域，并获得关键技能，为他们在学术界和工业界的前沿未来的职业生涯做好准备。新的研究生课程将得到本科生研究机会的补充。注重在线材料和资源将促进相互关联的学习和研究环境。外联工作将使公众参与讨论基因组工程和大规模计算的机遇和挑战。我们的学生将成为未来的领导者，在生物学的持续转变为定量学科，对基因组学如何影响人类生活的许多方面产生长期影响。

The complex developmental processes that enable an organism to differentiate a multitude of cell types from one genome are encoded in the intricacies of gene regulation, not the availability of a large repertoire of genes. The completion of the human genome has enabled us to decipher the processes that control gene regulation, from its context in three-dimensional chromosome folding to how specific transcription factors act, and finally, to its product as RNA or protein. Unraveling the multiscale genetic regulatory code requires experimental methods that are increasingly able to query regulatory function at genome-wide and high-throughput levels. A wide range of readouts can now be obtained from single cells, from in vitro effects of individual DNA sites to the in vivo impact of whole regulatory regions. Mechanistic readouts can be derived from genome editing that enables us to alter genomic sequence and to use “epigenetic” modifications to perturb the activity of regulatory regions.The resulting data create a need for sophisticated computational approaches to interpret the results and gain insights into cellular regulatory processes. Machine learning methods have seen a quantum leap in their ability to automatically infer higher-level representations from large complex data, e.g. in the context of deep learning. Combining experimental and computational approaches will provide an unprecedented opportunity to teach the next generation of researchers a quantitative understanding of genome function and gene regulation within the context of biological systems. The proposed alliance between Berlin institutions, led by Humboldt University, and Duke University will address this challenge through an international research training group (IRTG) composed of three complementary areas. (1) High-throughput genomics and editing, (2) bioinformatics and machine learning, and (3) developmental systems biology.Doctoral researchers will obtain a significant amount of their training at the partner institution, thereby benefitting from the synergy and expertise at both sites. They will be co-advised by computational and biological experts from both sides of the Atlantic throughout their training. A customized program will support the students in bridging the three emphasis areas and acquiring critical skills to prepare them for future careers at the cutting edge of academia and industry. New graduate courses will be complemented by research opportunities for undergraduates. A focus on online materials and resources will promote an interconnected learning and research environment. Outreach efforts will engage the public in the discussion of opportunities and challenges of genome engineering and large-scale computation. Our students will become future leaders in the ongoing transformation of biology into a quantitative discipline, with a long-term impact on how genomics affects numerous aspects of human life.