SGER: Sequence Alignment, Clustering, and Statistical Physics

0110903Yu这是SGER颁发的奖项，旨在支持生物信息学领域的理论研究和教育，使用不同于目前计算生物学所用的统计力学方法。研究的重点是序列比对和聚类。这一领域的一个重要问题是如何为从这些生物信息学工具获得的结果赋予准确的统计意义。PI最近开发的一种新的比对方法和统计理论将得到加强，以创建更完整的统计理论，并解决统计意义问题。这项工作的目的是开发一种更好的方法来提取适当的比对参数，其中插入、删除和替换生物单体将在平等的基础上处理。这项工作的另一个重点是使用随机聚类模型的变体来改进聚类方法。PI的新聚类方法假定不需要对现有数据的性质有先验知识，并有望对噪声数据更稳健。%此SGER奖支持生物信息学领域的理论研究和研究生教育。这项研究是跨学科的，目的是将统计物理学的强大工具用于理解已收集的海量基因组数据的意义这一更大的生物学问题的重要部分。PI的研究解决的问题是，能够可靠地识别不同序列之间的同源性，并改进可用于帮助识别与特定生物过程或疾病相关的基因表达模式的聚类工具。***

0110903YuThis is a SGER award to support theoretical research and education in the area of bioinfomatics using methods inspired by statistical mechanics that differ from ones currently used in computational biology. The research centers on sequence alignment and clustering. An important problem in this area is how to assign an accurate statistical significance to a result obtained from these bioinfomatics tools. A novel alignment method and statistical theory recently developed by the PI will be enhanced to create a more complete statistical theory and to address the statistical significance problem. An aim of this work is to develop a better method to extract appropriate alignment parameters where insertions, deletions, and substitutions of biomonomers will be treated on an equal footing. Another focus of this work is to improve clustering methods using variants of the random-cluster model. The PI's new clustering method assumes no a priori knowledge of the nature of the existing data and is expected to be more robust against noisy data.%%%This SGER award supports theoretical research and graduate level education in the area of bioinformatics. The research is interdisciplinary with a view to bring the formidable tools of statistical physics to bear on an important part of the larger biological problem of understanding the meaning of the enormous amount of genome data that has been collected. The PI's research addresses the problem of being able to reliably identify homology among different sequences and to improve clustering tools that can be used to help identify gene expression patterns associated with a certain biological process or disease. ***