Local ancestry inference for complex admixtures

摘要 所有的人类都是不同历史来源人群的混合体。这种混合物已经发生了 在一系列时间尺度上，从最近的混合，如 非洲裔美国人或西班牙裔美国人，到古代的混合体，如与尼安德特人的混合体 这发生在大约5万年前现代人迁出非洲的时候。 地方血统是一个人的染色体在每一个点的起源群体 基因组。当地血统对于许多应用是必不可少的，包括混合测绘和 推断人口统计历史。当地的血统不是直接观察到的，而是必须推断出来的 从一个人的基因数据。 现有的推断当地血统的方法不足以解开复杂的混合物 在人类群体中。当引用源的数据时，现有方法难以实现 种群数量有限或匹配不佳。这些方法不能在遗传上处理 相似的祖先群体，不同的混合时间，或者几个祖先 人口。 我们建议开发新的方法和计算工具来弥补这些差距。我们的 方法将利用最先进的单倍型频率模型和新的计算 极大地提高当地血统的精度和计算效率的方法 推论。我们的方法将通过对外加剂的灵活建模来克服目前的局限性 时间，当参考面板来自密切相关时，通过启用本地祖先推断 通过使用混合数据创建新的参考面板 参考种群与祖先种群匹配良好。我们的方法将是 在用户友好、计算高效的开源软件中实施，该软件可扩展到 对测序个体的大量样本进行分析。 我们将召唤来自不同非洲种群的细小规模的当地祖先的序列数据。我们会 使用当地祖先的电话来检测非洲过去的移民，来检测混血后的情况 选择，并对包括心脏在内的一系列性状进行混合作图， 肺部、血液特征，以及传染病情况。

Summary All humans are admixtures of various historical source populations. This admixture has occurred across a range of time-scales, from recent admixture such as the intercontinental admixture in African Americans or Hispanics, to ancient admixture such as the admixture with Neanderthals that occurred when modern humans migrated out of Africa around 50,000 years ago. Local ancestry is the population-of-origin of an individual’s chromosomes at each point in the genome. Local ancestry is essential for many applications, including admixture mapping and inferring demographic history. Local ancestry is not directly observed, but must be inferred from an individual’s genotype data. Existing methods for inferring local ancestry are inadequate for untangling complex admixtures in human populations. Existing methods struggle when reference data for the source populations are limited or poorly-matched. These methods are unable to handle genetically similar ancestral populations, divergent admixture times, or more than a few ancestral populations. We propose to develop new methods and computational tools to address these gaps. Our methods will utilize a state-of-the-art haplotype frequency model and new computational methods to greatly improve the accuracy and computational efficiency of local ancestry inference. Our methods will overcome current limitations by flexible modelling of admixture times, by enabling local ancestry inference when reference panels are from closely-related populations, and by creating new reference panels from admixed data when no existing reference population is well-matched to the ancestral population. Our methods will be implemented in user-friendly, computationally efficient, open-source software that scales to analysis of very large samples of sequenced individuals. We will call fine-scale local ancestry in sequence data from diverse African populations. We will use local ancestry calls to detect past migration within Africa, to detect post-admixture selection, and to perform admixture mapping for a broad spectrum of traits including heart, lung, and blood traits, and infectious disease status.