Deep learning approaches to decipher the impact of mobile element insertion on alternative splicing in neurological disorders

这项训练和研究应用的目的是利用深度学习技术的新发展来研究移动元素插入(MEI)在神经功能障碍(NDS)中的功能影响。MEI是可转座的DNA片段，能够插入整个人类基因组。至少有124种独立的MEI与人类疾病有关。这些疾病中约有20%代表了NDS的谱系，然而MEI对NDS病因的总体贡献尚未得到系统的估计。为了解决这一问题，我们将(1)确定健康个体GTEx队列中的功能MEI的特征；(2)构建MEI的全面功能图谱，以确定组织特异性和大脑特异性的影响；以及(3)将转录变化归因于各种NDS，其中将生成全基因组测序(WGS)数据。拟议中的应用还将为计算生物学家和统计遗传学家高达迪博士开发一个广泛的研究项目，他曾接受过神经退行性疾病替代剪接的功能基因组研究，以及针对导致严重神经发育障碍的剪接缺陷的治疗靶点。他开发了新的方法来研究转录组的调节，并促进药物开发中的分析。现在，他试图通过对大量来自对照和死后组织NDS病例的测序数据应用统计学和深度学习模型来扩展他的专业知识，然后在大规模疾病队列中归因于WGS的MEI的功能后果。培训计划包括两年的指导性研究，以学习基因组分析、梅特征描述和高级深度学习技术方面的新技能，然后是三年的独立实验室。该研究计划旨在通过分解针对MEI的转录变化来全面探索基因组中的功能变异。麻省总医院、哈佛大学和布罗德研究所的Michael Talkowski博士将担任主要导师，麻省理工学院和麻省理工学院计算生物学小组的Manolis Kellis博士将担任共同导师和密切合作伙伴。这些导师在基因组结构变异、功能基因组学、神经疾病的遗传学以及建立人类基因组中的功能元素的计算建模方面都是公认的专家。此外，一个由基础研究和翻译研究的独立调查人员组成的团队将为高晓松提供全面的反馈，以确保他的科学和职业发展都走上正轨。CGM、MGH、哈佛医学院、布罗德研究所和密歇根大学医学院的高度协作环境将为高博士向独立研究员的过渡做好准备。这个杰出的导师团队和培训项目将促进高博士的职业发展，因为他正在寻求重新定义人类基因组中MEI的功能图谱，并将它们在大规模神经疾病中的影响归因于它们。

The purpose of this training and research application is to study the functional impact of mobile element insertions (MEIs) in neurological disorders (NDs) using new developments in deep learning techniques. MEIs are transposable DNA fragments that are able to insert throughout the human genome. There are at least 124 independent MEIs associated with human diseases. Approximately 20% of these diseases represent a spectrum of NDs, yet the overall contribute of MEIs to the etiology of NDs has not been systematically estimated. To address this, we will (1) characterize functional MEIs in GTEx cohorts in healthy individuals; (2) build a comprehensive functional map of MEIs to determine tissue-specific and brain-specific impact; and (3) impute transcriptional changes on various NDs where whole-genome sequencing (WGS) data will be generated. The proposed application will also develop an extensive research program for Dr. Dadi Gao, a computational biologist and statistical geneticist who has trained in functional genomic studies of alternative splicing in neurodegenerative disorders and therapeutic targeting of a splicing defect that causes a severe neurodevelopmental disorder. He has developed novel methods to investigate regulation of the transcriptome and to facilitate analyses in drug development. He now seeks to expand his expertise by applying statistical and deep learning models on large cohorts of sequencing data from controls and cases with NDs from post-mortem tissues, then impute functional consequences of MEIs from WGS in large-scale disease cohorts. The training plan consists of two years of mentored research to learn new skills in genome analysis, MEI characterization, and advanced deep learning techniques, followed by three years of shaping an independent laboratory. The research plan is developed to comprehensively explore functional variation in the genome by decomposing transcriptomic changes against MEIs. Dr. Michael Talkowski at Massachusetts General Hospital, Harvard, and the Broad Institute will serve as the primary mentor, while Dr. Manolis Kellis at MIT and the MIT Computational Biology Group, and the Broad Institute will serve as a co-mentor and close collaborator. These mentors are recognized experts in genomic structural variants, functional genomics, the genetics of neurological disorders, and computational modeling to establish functional elements in the human genome. In addition, a team of independent investigators from basic and translational research will provide Dr. Gao with comprehensive feedback to keep both his science and career development on track. The highly collaborative environment in CGM, MGH, Harvard Medical School, the Broad Institute and the University of Michigan Medical School will prepare Dr. Gao for his transition to an independent investigator. This outstanding mentorship team and training program will facilitate the career development of Dr. Gao as he seeks to redefine the functional maps of MEIs in the human genome and to impute their impact in large-scale neurological disorders.