Training in Biomedical Discovery from Large Scale Data Sets

描述（由申请人提供）：基因组学、蛋白质组学和先进成像技术的发展，积累了大量的生物学数据。随着大规模数据集在生物医学研究中占据主导地位，我们正在接近一种范式转变，在这种转变中，发现的过程是数据驱动的，数据既是假设的来源，也是检验假设的手段。这些海量的数据是丰富的信息来源；然而，提取有意义的信息可能是一项艰巨的挑战，并且经常成为发现过程的瓶颈。因此，迫切需要对了解数据、数据是如何产生的以及数据的用途的科学家进行跨学科培训。此外，这些科学家必须成为分析大型数据集所需的新计算工具的开发人员和高技能用户。本课程的目标是培养学生在以下方面的精通：数据采集。这将包括基因组学、蛋白质组学和成像方法的知识。2. 计算。这将包括数学和统计算法的知识，有效的计算机代码的实施，以及强调在关系、演绎和其他数据库中的数据仓库方法。3. 数据集成。这是一个涉及从不同时空尺度的异构数据集中提取有用信息的关键领域。它将包括从分子到有机体水平的系统建模和模拟方法的知识。也将强调计算数据挖掘方法。该计划的核心将是跨学科团队的研究培训，在至少两名来自不同学科（即计算/数学和生物医学科学）的导师的指导下。培训活动将包括专门的教学课程、研讨会、期刊俱乐部和学生-教师静修活动。这将是一个跨机构的培训项目，教师来自计算机科学、统计学、遗传学和医学等部门，来自休斯顿地区墨西哥湾沿岸联盟（Gulf Coast Consortia）五个参与机构。培训有能力管理和从大型数据集中提取信息的科学家将极大地促进传染病和癌症等领域的生物发现，因此这一培训方案将对公共卫生产生直接的积极影响。

DESCRIPTION (provided by applicant): The development of genomics, proteomics and advanced imaging technology has resulted in the accumulation of vast amounts of biological data. As large scale data sets become predominant in biomedical research, we are approaching a paradigm shift in which the process of discovery is data-driven, and in which data are the source of hypotheses as well as the means for testing them. These masses of data are rich sources of information; however, extracting meaningful information can be a daunting challenge, and often presents a bottleneck for the discovery process. Thus, there is a pressing need for interdisciplinary training of scientists who understand the data, how they are generated, and what they are used for. In addition, these scientists must become developers and highly skilled users of the new computational tools necessary to analyze large data sets. The goal of this program is to train students to become proficient in the following areas: 1. Data acquisition. This will include knowledge of the methods of genomics, proteomics and imaging. 2. Computation. This will include knowledge of mathematical and statistical algorithms, implementation of effective computer codes as well as an emphasis on methods of data warehousing in relational, deductive and other databases. 3. Data integration. This is a critical area that involves extracting useful information from the heterogeneous data sets at various spatial and temporal scales. It will include knowledge of methods of modeling and simulation of systems from the molecular to the organism level. There will also be an emphasis on computational data mining methods. The core of the program will be research-based training in interdisciplinary teams under the guidance of at least two mentors from disparate disciplines (i.e., computational/mathematical and biomedical sciences). Training activities will consist of specialized didactic coursework as well as seminars, journal clubs and a student-faculty retreat. This will be a cross-institutional training program with faculty drawn from departments ranging from computer science and statistics to genetics and medicine, in five participating institutions in the Gulf Coast Consortia in the Houston Area. The training of scientists equipped to manage and extract information from large data sets will greatly facilitate biological discovery in areas such as infectious disease and cancer and therefore this training program will have a direct, positive impact on public health.