Detecting Disease-Associated Genotype Patterns
Background: In addition to single-locus (main) effects of disease variants, there is a growing consensus that gene-gene and gene-environment interactions may play important roles in disease etiology. However, for the very large numbers of genetic markers currently in use, it has proven difficult to develop suitable and efficient approaches for detecting effects other than main effects due to single variants.Results: We developed a method for jointly detecting disease-causing single-locus effects and gene-gene interactions. Our method is based on finding differences of genotype pattern frequencies between case and control individuals. Those single-nucleotide polymorphism markers with largest single-locus association test statistics are included in a pattern. For a logistic regression model comprising three disease variants exerting main and epistatic interaction effects, we demonstrate that our method is vastly superior to the traditional approach of looking for single-locus effects. In addition, our method is suitable for estimating the number of disease variants in a dataset. We successfully apply our approach to data on Parkinson Disease and heroin addiction.Conclusions: Our approach is suitable and powerful for detecting disease susceptibility variants with potentially small main effects and strong interaction effects. It can be applied to large numbers of genetic markers.
Quan Long Qingrun Zhang Jurg Ott
Beijing Institute of Genomics, Chinese Academy of Sciences, No.7 Bei Tu Cheng West Road, Beijing 100029, China
国际会议
The 7th Asia-Pacific Bioinformatics Conference(第七届亚太生物信息学大会)
北京
英文
781-787
2009-01-01（万方平台首次上网日期，不代表论文的发表时间）