Understanding Transmission with Integrated Genetic and Epidemiologic Inference

这个项目的总体目标是开发和验证新的方法来执行联合推理， 结合流行病学和遗传学数据。这一推理方法旨在提供 基本传输参数，如RO，以及提供未观察到的传输估计 树木和未观察到的人口中的易感，感染和恢复的个人， 时间我们关注两种常见的情况。首先，我们的目标是密集采样，但局部， 流行病学和遗传学数据，其中人、地点和时间已知，病原体 获得遗传样本。这类数据集通常在传输研究期间生成 在家庭、学校和类似环境中，以及在SARS或H7N9等新型疫情的分析中， 我们的推理框架旨在从流行病学和免疫学的结合来估计宿主到宿主的传播网络。 和基因数据。在第二种情况下，我们的目标是抽样稀疏，但范围更广，流行病学 和遗传数据，在这些数据中，我们观察了病例报告的时间序列和稀疏采样的病原体遗传数据， 序列的在这个推理框架中，我们试图从一个 相对较小的案例。该框架利用聚结理论从采样的 基因序列到种群水平的动态。在实现中，我们计划利用复杂的推理， 结合马尔可夫链蒙特卡罗（MCMC）和顺序蒙特卡罗（SMC）的方法 粒子MCMC（Particle MCMC）我们计划利用这些新的推理方法， 研究流感传播异质性和局部传播结构， 如果没有流行病学和遗传学的综合推理框架，很难进行全面分析。

The overall goal of this project is to develop and validate novel methods to perform joint inference from combined epidemiologic and genetic data. This inference methodology seeks to provide estimates of fundamental transmission parameters, such as RO, as well as provide estimates of unobserved transmission trees and unobserved counts of susceptible, infected and recovered individuals in the population through time. We focus on two common scenarios. In the first, we target densely sampled, but localized, epidemiologic and genetic data, in which the person, place and time are known, and in which pathogen genetic samples are obtained. These sorts of datasets are commonly generated during transmission studies in households, schools, and similar settings, but also in analyses of novel outbreaks such as SARS or H7N9. Our inference framework seeks to estimate host-to-host transmission networks from combined epidemiologic and genetic data. In the second scenario, we target sparsely sampled, but broader in scope, epidemiologic and genetic data, in which we observe a time series of case reports and sparsely sampled pathogen genetic sequences. In this inference framework, we seek to model population-level transmission processes from a relatively small samples of cases. This framework utilizes coalescent theory to extrapolate from sampled genetic sequences to population-level dynamics. In implementation, we plan to utilize sophisticated inference methodology that combines Markov chain Monte Carlo (MCMC) and sequential Monte Carlo (SMC) approaches in what's termed particle MCMC (PMCMC). We plan to utilize these novel inference methods to investigate transmission heterogeneity and local transmission structure in influenza, phenomena that have been difficult to fully analyze without a combined epidemiologic and genetic inference framework in place.