Mathematically Rigorous Results In Sequence Matching.

序列匹配中严格的数学结果。

• 批准号: 10007523
• 负责人: John Spouge
• 金额: $ 2.38万
• 依托单位: NATIONAL LIBRARY OF MEDICINE
• 依托单位国家: 美国
• 资助国家: 美国
• 起止时间: 至
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10007523
• 关键词: Architecture; Benchmarking; Data; Databases; Detection; Disease; Genetic Diseases; Joints; Malignant Neoplasms; Maryland; Mathematics; Measures; Methods; Mutation; Proteins; Publishing; Statistical Methods; Testing; Universities; Work; programs; protein function; simulation; statistics

Dr Maricel Kann and her group at the University of Maryland had a list of mutations in disease, but no statistical test to decide whether the mutations were harmless or disease-causing. Joint work with the Drs DoHwan and Junyong Park in the Statistics Department at the University of Maryland established the validity of a proposed test with extensive simulations. The test used the asymptotic regression we developed for the estimation of BLAST statistical parameters, and the simulation showed that the method was more powerful than methods using Efron's local false-discovery rate. Dr Kann published work applying the statistical test to the mutational data. We also developed a benchmarking domain database, MultiDomainBenchmark, to evaluate the results of programs that search for specific domain architectures, corresponding to proteins with specific functions.

马里兰州大学的Maricel Kann博士和她的研究小组有一份致病突变的清单，但没有统计学检验来确定这些突变是无害的还是致病的。与马里兰州大学统计系的DoHwan博士和Junyong Park博士的联合工作通过广泛的模拟确定了拟议测试的有效性。该测试使用的渐近回归，我们开发的BLAST统计参数的估计，和模拟表明，该方法是更强大的方法，使用Efron的本地错误发现率。Kann博士发表了将统计检验应用于突变数据的工作。 我们还开发了一个基准域数据库，MultiDomainBenchmark，以评估搜索特定域架构的程序的结果，对应于具有特定功能的蛋白质。