A CLASSICAL LIKELIHOOD APPROACH FOR ADMIXTURE MAPPING USING THE EM ALGORITHM

这个子项目是许多研究子项目中利用 资源由NIH/NCRR资助的中心拨款提供。子项目和 调查员(PI)可能从NIH的另一个来源获得了主要资金， 并因此可以在其他清晰的条目中表示。列出的机构是 该中心不一定是调查人员的机构。 最近提出了几种疾病地图绘制方法，这些方法使用了来自历史上不同地理来源的最近混合的人口所产生的信息。这些方法包括经典的似然方法和贝叶斯方法。在这项研究中，我们使用EM算法直接最大化混合映射的隐马尔可夫模型的似然函数，允许模型参数的不确定性，例如亲本群体中的等位基因频率。我们通过检查祖先等位基因频率估计和个体标记-位置特定祖先来确定该方法的稳健性，当数据由不同的群体混合模型产生并且不使用学习样本时。对于不同的人口混合模型所产生的数据，所提出的方法优于广泛使用的贝叶斯MCMC策略。祖先的多点信息内容是根据Smith等人提供的地图推导出来的。(2004)，并计算了相关的统计功率。还考虑了祖先等位基因频率的错误指定和相邻标记之间的连锁不平衡的影响。最后，我们检查了混合LD在基因组中的分布，包括真实数据和模拟数据，并建立了适用于混合作图研究的全基因组意义的阈值。进行混合作图的软件ADMIXPROGRAM可从作者处获得。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Several disease-mapping methods have been proposed recently, which use the information generated by recent admixture of populations from historically distinct geographic origins. These methods include both classic likelihood and Bayesian approaches. In this study we directly maximize the likelihood function from the hidden Markov Model for admixture mapping using the EM algorithm, allowing for uncertainty in model parameters, such as the allele frequencies in the parental populations. We determined the robustness of the proposed method by examining the ancestral allele frequency estimate and individual marker-location specific ancestry when the data were generated by different population admixture models and no learning sample was used. The proposed method outperforms a widely used Bayesian MCMC strategy for data generated from various population admixture models. The multipoint information content for ancestry was derived based on the map provided by Smith et al. (2004) and the associated statistical power was calculated. The effects of misspecification of ancestral allele frequencies and linkage disequilibrium between adjacent markers were also considered. Finally, we examined the distribution of admixture LD across the genome for both real and simulated data and established a threshold for genome wide significance applicable to admixture mapping studies. The software ADMIXPROGRAM for performing admixture mapping is available from the authors.