CAREER: Improving the Performance of Motif Finding Tools through Novel Reliable Significance Estimation and a Study of DNA Replication Origins

康奈尔大学的一名研究人员因发现DNA序列中的调控基序而获得了美国国家科学基金会的职业资助。它仍然是计算生物学中的一个基本问题，因为识别这些调控元件是理解基因表达调控的核心。最近一项广泛的比较研究表明，在检测实际结合位点时，现有工具有很大的改进潜力。该项目的第一个目标是开发一个可靠的显著性分析的轮廓为基础的从头基序查找器。这种分析可以进一步帮助描述这些寻找器的理论极限，并将它们的性能推向极限。在过去的几年里，出现了几种新类型的数据，这些数据被成功地集成到motif finder中。特别是，随着密切相关物种的可用性增加，一类具有系统发育意识的基序查找器已经开发出来。第二个主要目标是开发一种新的有效的显著性分析，可用于分析集成系统发育或定位数据的基序查找器的结果，从而提高它们的性能。第三个目标是通过与分子生物学家的合作，更好地表征酵母物种的复制起点，特别是表征酵母中复制起点可变性的序列元素，以及检测和分析相关物种的新复制起点。该提案的教育部分将培养和丰富计算机科学、生物学和统计学各个学术水平的学生，让他们接触到这三个学科，这三个学科本质上结合在一起定义了计算生物学。

A Cornell University researcher is awarded an NSF CAREER grant to uncover regulatory motifs in DNA sequences. It remains a fundamental problem in computational biology as identification of such regulatory elements is central to understanding regulation of gene expression. A recent extensive comparative study showed there is potential for great improvement in existing tools when it comes to detection of real binding sites. The first goal of this project is to develop a reliable significance analysis for profile based de novo motif finders. This analysis can then further assist in delineating the theoretical limit of these finders and in pushing their performance envelope toward that limit. Over the last few years several new types of data emerged that were successfully integrated into motif finders. In particular, with the increased availability of closely related species, a class of phylogeny-aware motif finders has been developed. The second major goal is to develop a new efficient significance analysis that can be used to analyze the results of motif finders that integrate phylogeny or localization data and thereby improve their performance. The third goal is to acquire in a collaborative work with a molecular biologist a better characterization of replication origins in yeast species, in particular, characterizing the sequence elements that account for the variability among replication origins in yeast as well as in detecting and analyzing new replication origins in related species. An education component of this proposal will train and enrich students of all academic levels from Computer Science, Biology and Statistics exposing each one to all three disciplines that essentially combine to define computational biology.