An Outlier Detection Method for Scanning Disease-Causing Genes

• Main Authors: Chen-Chi Tien, 田珍綺
• Other Authors: John Jen Tai
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/80197883804041152825

碩士 === 輔仁大學 === 應用統計學研究所 === 95 === The study of locating genes of underlying susceptibility to complex disease is a challenging topic in genetic analysis. A complex disease is usually affected by both the environments and multiple loci. Moreover, it may also be influenced by the interactions between disease genes and the interactions between genes and environments. Simultaneous association test using multiple markers can detect the possible gene locations for complex diseases. In this thesis we propose a new statistical method to analyze the multiple-gene data. This method uses the multiple P values from the single locus tests as a base to produce a combined statistic. Then by omitting a marker and using the rest markers another combined statistic can be calculated. Taking the difference between the two combined statistics we have the new statistic for detecting the effect of the omitted locus. Using this outlier detection idea the scan process can be conducted. Simulation studies showed that power of the proposed multilocus association test significantly outperforms than single locus association test when one disease-causing gene exists. This study is the preliminary study to examine whether the outlier detection idea is applicable in genome-wide scan. There still are many unclear points that need future works.