CAREER: Models and Algorithms for Comparative Analysis of Biological Networks

最近出现的用于测量分子相互作用的高通量技术已经产生了大量的生物网络，这使得能够对复杂的生物有机体进行系统的研究。由于具有关键功能的生物途径通常在不同的生物体中是保守的，因此对这些网络进行比较分析可以提供一种很好的方法来研究生物网络的组织，以及追踪新的途径并研究它们的功能。智力优势：该项目旨在为比较网络分析开发一个坚实的数学框架，并为比较基因组规模的生物网络设计创新技术。主要研究目标包括：（1）建立一个多状态半马尔可夫随机游走（SMRW）模型，用于大规模网络的概率比较;（2）开发高效的生物网络查询和比对算法;（3）应用所开发的算法识别新的生物通路，并研究其功能和详细机制。该项目的研究活动与全面的教育计划紧密结合，其总体目标在于提高学生的能力。通过研究和教育的整合学习经验。这一目标转化为三部分的教育计划：（1）开发一个概念库存（CI）的基因组信号处理和计算生物学（称为CIGSP）;（2）转移研究生水平的课程？概率图模型？以问题为基础的格式;（3）基于成熟和新兴的教学方法，设计一门新的网络生物学概率模型本科课程。新的概念库存CIGSP将提供一个有价值的诊断/评估工具，以加强在基因组信号处理和计算生物学的教育，它将被用于PI？的课程，设计可衡量的教育目标和评估学习成果。

Recent advent of high-throughput technologies for measuring molecular interactions has yielded large collections of biological networks, which enable systematic studies of complex biological organisms. Since biological pathways with critical functions are often conserved across different organisms, comparative analysis of these networks can provide an excellent way of investigating the organization of biological networks, as well as tracking down novel pathways and studying their functions. Intellectual Merit: This project aims to develop a solid mathematical framework for comparative network analysis and devise innovative techniques for comparing genome-scale biological networks. The main research objectives include: (1) develop a multi-state semi-Markov random walk (SMRW) model for probabilistic comparison of large-scale networks; (2) develop efficient algorithms for querying and aligning biological networks; (3) apply the developed algorithms to identify novel biological pathways and investigate their functions and their detailed mechanisms.Broader Impact: The research activities in this project are closely integrated with a comprehensive educational plan, whose overall goal lies in enhancing students? learning experience through the integration of research and education. This goal is translated into a three-part educational plan: (1) develop a concept inventory (CI) for genomic signal processing and computational biology (called the CIGSP); (2) shift a graduate-level course on ?Probabilistic Graphical Models? to a problem-based format; (3) design a new undergraduate course on Probabilistic Models for Network Biology based on proven and emerging pedagogical approaches. The new concept inventory CIGSP will provide a valuable diagnosis/assessment tool for enhancing education in genomic signal processing and computational biology, and it will be used in the PI?s courses, to design measurable educational objectives and evaluate the learning outcomes.