COBRE: UID: PROJ 4: STATISTICAL METHODS FOR THE ANALYSIS OF MICROBIAL COMMUNITY

这个子项目是许多研究子项目中的一个 由NIH/NCRR资助的中心赠款提供的资源。子项目和 研究者（PI）可能从另一个NIH来源获得了主要资金， 因此可以在其他CRISP条目中表示。所列机构为 研究中心，而研究中心不一定是研究者所在的机构。 疾病的发生不仅是由于孤立的有害病原体，而且也是由于人类-微生物生态系统平衡的破坏。这一新出现的知识需要修订目前的诊断方法，以纳入有关这些生态系统的“正常”状态以及偏离这种状态可能导致疾病的性质的信息。它还需要使这些修订后的方法随时可供医学界使用，以方便诊断疾病。该项目的目标是开发计算方法来区分正常和异常（与疾病症状相关）的微生物群落，同时考虑元数据（如性别，年龄，种族和健康史）。具体而言，我们的目标是：1）开发基于模型的聚类方法，以准确区分和表征不同的微生物群落组，以及2）开发基于模型的分类方法，以正确和有效地将新采样的微生物群落分类为预先存在的、良好表征的组。我们将在三个阶段实现这些具体目标：建模，实施和验证。建模将涉及使用概率理论和贝叶斯框架来获取数据中的可用信息。实施将涉及在模型参数估计和微生物群落群关联中使用蒙特卡罗方法。将使用真实的和模拟数据验证所提出方法的准确性。我们的目标的实现将形成一个决策支持系统的基础，以帮助临床医生确定偏离正常的微生物群落结构。为此，我们将开发软件，供医学界用于研究和诊断目的。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Diseases occur not only due to harmful pathogens that act in isolation but also due to disruption of the balance of the human-microbial ecosystem. This emerging knowledge requires a revision of the current diagnostic approaches to incorporate information about the "normal" state of these ecosystems and the nature of deviation from this state that can result in disease. It also requires making these revised approaches readily available to the medical community to facilitate diagnosing diseases. The goal for this project is to develop computational approaches to differentiate between normal and abnormal (associated with disease symptoms) microbial communities, taking metadata (such as gender, age, ethnicity, and health history) into consideration. Specifically, we aim to: 1) develop model-based clustering methods to accurately distinguish and characterize different microbial community groups and 2) develop model-based classification methods to correctly and efficiently classify newly sampled microbial communities to pre-existing, well characterized groups. We will attain these specific aims in three stages: modeling, implementation and validation. Modeling will involve the use of probability theory and a Bayesian framework to capture the information available in the data. Implementation will involve the use of Monte Carlo methods in model-parameter estimation and microbial community-group association. Validation of the accuracy of the proposed methods will be performed using real and simulated data. Achievement of our Aims will form the basis of a decision-support system to assist clinicians in identifying deviations from normal microbial community structures. For this purpose we will produce software that will be made available to the medical community for research and diagnostic purposes.