EAGER: A novel set of computational methods for mining nonlinear and high-order relationships

研究网络数据集中的非线性和高阶关系对于理解复杂系统的奥秘和人体微生物群落、环境生态系统等真实的世界问题的结构具有重要意义。然而，庞大的数据量、复杂性和错综复杂的数据属性给数据分析和挖掘带来了机遇和挑战。该项目旨在开发一种新的计算框架来解决这些具有挑战性的问题，重点关注以下两项任务：1）挖掘，提取和推断相互作用和关系的新计算方法; 2）从三种类型的微生物组数据集识别高阶相互作用和关系的新计算方法：宏基因组，细菌基因组和文献。该研究具有高风险和高回报，因为其结果将彻底改变构建和分析微生物知识图的方式，并有助于发现生物机制和医学应用。本项目将考虑微生物组学数据的特点，并开发一种新的微生物组学数据分析计算框架。可扩展的概率和张量方法与流形正则化挖掘微生物组学数据将克服线性，欧氏和无限空间的假设。这些计算方法将用于构建和分析微生物知识图，以帮助发现。计算结果将通过开放式软件工具（通过开发一个新的R软件包）以及在会议和研讨会上的演示来传播。该提案的研究和教育计划都是高度跨学科的，吸引了来自各个研究领域的学生和教师，并借鉴了多个研究领域的工作。拟议的研究领域有助于提高学生在多个层面的科学好奇心。学生将获得显着接触大数据，计算科学，生物信息学和统计学的最新研究。

Studying nonlinear and high-order relationships in a network data set is very important to understand the mystery of a complicated system and the structure of many real world problems such as the microbial community of human body, environmental eco-systems, etc. However, the huge data volume, the complexity and the intricate data properties have created a lot of opportunities and challenges for data analysis and mining. This project aims to develop a novel computational framework to tackle these challenging issues, focusing on the following two tasks: 1) Novel computational approaches to mine, extract and infer interactions and relations; 2) Novel computational methods for identifying higher-ordered interactions and relations from three types of microbiome datasets: metagenomes, bacterial genomes and literature. This research is of high risk and high payoff because the outcome will revolutionize the way to construct and analyze microbial knowledge graphs, and to aid discovery for biological mechanisms and medical applications.This project will consider the characteristics of microbiomic data and develop a novel computational framework for microbiomic data analysis. Scalable probabilistic and tensor methods with manifold-regularization for mining microbiomic data will overcome the assumption of linear, Euclidean and infinite space. These computational methods will be used to construct and analyze microbial knowledge graphs to aid discovery. The computational results will be disseminated through open software tool (by developing a novel R package) and presentations at conferences and workshops. Both the research and education plans of the proposal are highly interdisciplinary, engaging students and faculty from various research areas and drawing from work on multiple fields of study. The proposed research area lends itself to raising the scientific curiosity of students at many levels. Students will obtain significant exposure to the latest research in big data, computational science, bioinformatics and statistics.