Predictive Modeling with Clinical and Genomic Data in COPD

描述（由申请人提供）：候选人：Peter Castaldi博士是一名医生，正在完成f32资助的一段时间。2009年7月1日，他将在塔夫茨医学中心和临床研究与卫生政策研究所（ICRHPS）开始全职工作。该职位有25%的临床承诺。他的主要研究兴趣是慢性阻塞性肺病的遗传流行病学以及将基因组发现转化为临床实践和公共卫生。他的特别兴趣是遗传元分析，基因-环境相互作用，以及基于回归和机器学习方法的预测建模。他的近期目标是：1.)通过多个全基因组关联（GWA）研究的联合分析，确定与COPD易感性和COPD相关表型的新遗传关联；2.)确定遗传和吸烟基因的相互作用；3.)利用临床和基因组信息，开发慢性阻塞性肺疾病（COPD）的准确预测模型。他的长期目标是成为一名具有生物信息学专长的独立研究者。他对实现这一目标的愿景包括发展生物信息学方面的专业知识，以便能够参与并最终领导多学科团队，将计算方法应用于基因组数据集，以回答将改善患者护理和人口健康的重要临床问题。环境：Castaldi博士将在丰富的跨学科环境中接受培训。他的主要导师Joseph Lau博士是ICRHPS临床证据综合中心的负责人，他是荟萃分析和证据综合领域的全球领导者。在塔夫茨大学，除了与刘博士定期会面外，彼得还将接受该领域领导者的遗传证据合成培训，包括刘博士。约翰·约阿尼迪斯和汤姆·特里卡里诺斯。该应用程序的联合导师Edwin Silverman博士是钱宁实验室COPD遗传学的主要研究员。在钱宁实验室，彼得将接受呼吸遗传学和遗传流行病学方面的优秀培训，他将拥有最先进的高通量基因分型、下一代测序技术和生物信息学支持的资源。Castaldi博士还将继续与塔夫茨大学计算机科学系的Donna Slonim博士合作，后者将为计算算法在基因组数据中的应用提供帮助，并为Castaldi博士继续建立生物信息学的实践和理论基础提供指导。研究：慢性阻塞性肺病是发病率和死亡率的主要原因，对公共卫生的重要性日益增加。虽然吸烟是慢性阻塞性肺病的主要危险因素，但一般人群对吸烟引起的肺损伤的易感性各不相同。有强有力的证据支持遗传因素对COPD易感性的影响。了解基因和环境如何相互作用产生临床COPD将允许更准确的诊断工具，并为COPD治疗的发展开辟新的研究途径。我们建议：1.)通过对4项大型COPD GWA研究的患者水平数据进行荟萃分析，确定COPD易感性和4种COPD相关表型的新遗传关联；2.)确定吸烟基因和基因相互作用；3.)建立COPD易感性和COPD相关表型的预测模型。为了最大限度地从基因组数据中获得信息，我们将结合多个研究的数据（国家肺气肿治疗试验遗传学辅助研究，挪威病例对照研究，COPDGene和ECLIPSE -总样本量=7,962）来增加功率，并采用基于回归和机器学习的方法来识别基因型数据中相互作用的复杂模式。我们的研究旨在探索并随后严格验证发现的主要效应和相互作用关联。使用预测模型，我们将量化包括遗传主效应和遗传相互作用数据到传统临床变量的增量预测效益。相关性：拟议的工作将确定与COPD相关的新基因，并将其置于多变量背景下，以便更好地了解遗传差异和环境暴露如何促进COPD的发展。这项工作产生的模型将有助于将基因组发现转化为临床实践和公共卫生，与NHLBI的使命保持一致，即促进心脏、肺和血液疾病的预防和治疗，增强所有人的健康，使他们能够活得更长、更充实。

DESCRIPTION (provided by applicant): Candidate: Dr. Peter Castaldi is a physician completing a period of F32-funded support. On July 1st, 2009 he will begin a full-time position at Tufts Medical Center and the Institute for Clinical Research and Health Policy Studies (ICRHPS). This position involves a 25% clinical commitment. His principal research interests are the genetic epidemiology of COPD and the translation of genomic discoveries into clinical practice and public health. His particular interests are genetic meta-analysis, gene-environment interaction, and predictive modeling with regression-based and machine-learning methods. His immediate goals are 1.) to identify novel genetic associations with COPD susceptibility and COPD- related phenotypes through the combined analysis of multiple genome-wide association (GWA) studies, 2.) to identify epistatic and gene-by-smoking interactions, and 3.) to develop accurate predictive models in chronic obstructive pulmonary disease (COPD) using clinical and genomic information. His long-term goal is to be an independent investigator with expertise in bioinformatics. His vision for achieving this goal involves developing expertise in bioinformatics so as to be able to participate in and eventually lead multidisciplinary teams in the application of computational methods to genomic datasets in order to answer important clinical questions that will improve the care of patients and population health. Environment: Dr. Castaldi will receive training in a rich, interdisciplinary environment. His principal mentor, Dr. Joseph Lau, is the head of the Center for Clinical Evidence Synthesis in the ICRHPS, and he is a worldwide leader in the field of meta-analysis and evidence synthesis. At Tufts, in addition to regular meetings with Dr. Lau, Peter will receive training in genetic evidence synthesis from leaders in the field, including Drs. John Ioannidis and Tom Trikalinos. The co-mentor of this application, Dr. Edwin Silverman, is a leading researcher in COPD genetics at the Channing Laboratory. At the Channing Laboratory, Peter will receive excellent training in respiratory genetics and genetic epidemiology, and he will have resources to state of the art high-throughput genotyping, next-generation sequencing technologies, and bioinformatics support. Dr. Castaldi will also continue his collaboration with Dr. Donna Slonim in the Tufts Computer Science Department, who will provide assistance with application of computational algorithms to genomic data and guidance as Dr. Castaldi continues to build a practical and theoretical foundation in Bioinformatics. Research: COPD is a major cause of morbidity and mortality that is of increasing public health importance. While the principal risk factor for COPD, smoking, is well-established, there is variable susceptibility in the general population to the lung damage caused by cigarette smoke. There is strong evidence supporting a genetic component to COPD susceptibility. Understanding how genes and environment interact to produce clinical COPD will allow for more accurate diagnostic tools and open new avenues of investigation for the development of COPD therapies. We propose to 1.) identify novel genetic associations with COPD susceptibility and 4 COPD-related phenotypes by performing meta-analysis on patient-level data from 4 large COPD GWA studies, 2.) identify gene-by-smoking and gene-gene interactions, and 3.) develop predictive models for COPD susceptibility and COPD-related phenotypes. In order to maximize the information obtained from genomic data, we will combine data from multiple studies (the National Emphysema Treatment Trial Genetics Ancillary Study, the Norway Case-Control Study, COPDGene, and ECLIPSE - total sample size=7,962) to increase power and employ regression-based and machine-learning methods to identify complex patterns of interaction in genotype data. Our study is designed to both explore and subsequently rigorously validate discovered main effect and interaction associations. Using predictive models, we will quantify the incremental predictive benefit of including genetic main effects and genetic interaction data to traditional clinical variables. Relevance: The proposed work will identify new genes associated with COPD and place them in a multivariate context so as to develop a better understanding of how genetic differences and environmental exposures contribute to the development of COPD. The models generated by this work will facilitate the translation of genomic discoveries to clinical practice and public health, in keeping with the NHLBI's mission to promote the prevention and treatment of heart, lung, and blood diseases and enhance the health of all individuals so that they can live longer and more fulfilling lives. PUBLIC HEALTH RELEVANCE: The proposed work will identify new genes associated with COPD and place them in a multivariate context so as to develop a better understanding of how genetic differences and environmental exposures contribute to the development of COPD. The models generated by this work will facilitate the translation of genomic discoveries to clinical practice and public health, in keeping with the NHLBI's mission to promote the prevention and treatment of heart, lung, and blood diseases and enhance the health of all individuals so that they can live longer and more fulfilling lives.