面向阿尔茨海默症的多维度多组学数据融合挖掘方法研究

探索阿尔茨海默症（AD）的多层次发病机制有利于疾病的诊断和预防，同时也是老龄化社会急需研究的重要课题。本项目基于不同组学数据之间的互补性特点，对病变脑区、风险基因和异常蛋白等发病因素进行整合，形成宏观—微观—宏观的研究回路，深入研究AD的多层次发病机制：基于影像组数据提出脑网络动态分析方法，从宏观层面揭示AD不同发病阶段的脑网络演变模式；融合影像组和基因组数据，从宏观到微观挖掘AD脑网络演变与风险基因的关联模式；利用蛋白组数据构建风险基因相关的蛋白质网络，设计异质网络分析技术，从微观到宏观探究AD发病过程中蛋白质网络对脑网络的调控模式；设计大数据存储与分析框架，完善多组学数据融合分析模型与算法，基于不同高性能平台设计数据划分与调度策略，达到数据、算法、平台的最优组合。本研究有助于发现AD新的药物靶点，提供特色化治疗方案，推动数字健康和精准医疗技术的发展。

Exploring the multi-level pathogenesis of Alzheimer’s disease (AD) is conducive to the diagnosis and prevention of the disease, and it is also an important topic in the aging society. Based on the complementary characteristics of different omics data, this project integrates multiple morbigenous factors including diseased brain regions, risk genes and abnormal proteins to form a macroscopic-microcosmic-macroscopic research loop, and to further delve into the multi-level pathogenesis of AD. For the neuroimaging data, the dynamic analysis technology of brain network is designed to reveal the evolution pattern of brain network in different stages of AD from the macroscopic level. The neuroimaging and genomic data are fused to detect the association pattern between diseased brain regions and risk genes in AD from macroscopic to microcosmic level. The proteome data is used to construct the protein network related with risk genes, and the heterogeneous network analysis technology is designed to explore the regulation pattern of protein network on brain network in the process of AD from microcosmic to macroscopic level. The data storage and analysis framework is designed, and the multi-omics data fusion models and algorithms are improved. Based on the different high-performance platform, the efficient data partitioning and scheduling strategy is designed to achieve the optimal combination of data, algorithms and platforms. This study is helpful to find new drug targets, provide characteristic treatment programs for AD, and promote the development of digital health and precision medical technology.