Genotype-based association analysis via entropy

With the advent of large-scale genotyping technologies, enormous quantities of genotype data that were generated have been well exploited through phased haplotypes, and the haplotype-based association study is used as one of the major statistical methods for gene mapping of human complex traits. However, haplotype-based method depends on the information of haplotype frequencies, which results in infeasible computation when haplotypes are not directly observed. This paper provides a genotype-based statistic with multiple tightly linked markers for association analysis using entropy theory. The statistic here does not require haplotype phasing and only requires genotype data. The distribution and the power of the statistic are investigated by simulative study. The results show that the statistic has very reasonable performance. We demonstrated the powerfulness of the statistic by applying our approach to a specific example on hereditary hemochromatosis.