基于数量性状的纯合子定位分析方法研究

Homozygosity mapping (HM) is the most effective for recessive (qualitative) traits, and lengths of runs of homozygosity (ROHs) are often compared between case and control individuals. Here we propose to extend this approach to quantitative trait loci (QTL) underlying qualitative traits (QT).This has rarely, if ever, been done for QTLs except by dichotomizing the QT into an “affected” and a “normal” range to which classical HM is then applied. QTs occur in many areas of human and animal genetics.We will develop novel approaches to HM in QTs, implement them in computer programs, and make resulting software freely available to researchers. Here is a list of specific aims to be addressed in this grant proposal..1)To collect published QT data and to document that ROHs length, L, varies with the QT value, Q. We use one of the generally available HM approaches to generate ROHs with default parameter settings like maximum number of heterozygous and missing genotypes allowed within an ROH. Thus, for each Q value (individual), there will generally be several ROHs. If the number of ROHs per individual is small then we will form groups of individuals. For each Q value or groups of Q values, we will compute the average length L of ROHs to see how these averages change with Q values..2)Proportion of long ROHs. Because only large L values tend to be genetically relevant we will define a threshold t and distinguish large values, L > t, from moderate to small values, L < t, and work with the proportion of large L values associated with each Q value (individual). This dependency may be tested by chi-square whose total value may be broken down into two components, a linear increase in proportions (with 1 df) and a remainder. A refinement of this approach is to assume that a linear increase occurs only at one or the other end of the distribution of Q values and that for the remaining Q values the proportion of large L values will be constant. We will fit a simple spline consisting of two straight lines, a horizontal line for most Q values and a sloping line for one of the ends of the Q distribution..3)Parameter estimation. Some of the models we propose will not be manageable with standard statistical tests. For example, the spline consisting of two straight lines mentioned in aim 2 has a small number of parameters that need to be estimated by ad hoc methods. We plan to apply least squares (or maximum likelihood) estimation with numerical maximization of parameter values..4)Smallest regions of overlap. In classical HM on QT, what is often done is to match long ROHs for several affected individuals. For example, one may determine the smallest region of overlap (SRO) in long ROHs for 80% of the affected individuals. We will extend this approach to QTLs by forming groups of individuals with similar QT values. We can then proceed in analogy to approaches considered above, for example, compute averages of SROs and determine how these averages change with QT values.

纯合子定位法是最常见的稀有常染色体隐性遗传病致病基因精确定位方法，到目前为止该方法仅适用于分析病例对照性状。越来越多的研究表明许多复杂性状是由稀有变异造成的，而很多的复杂性状都是数量性状，所以数量性状位点的定位在复杂性状的遗传分析中占有重要地位。申请人前期研究已发现：如果数量性状遗传因素至少一部分由隐性基因控制，那么在不同的数量性状表型分类组中连续性纯合片段(ROHs)长度分布是不同的。基于此，本研究拟建立一系列遗传统计方法来寻找、解释和检定ROHs长度与数量性状关系，计算群体中ROHs重叠的最小区域的遗传效应和ROHs的参数估计及预测，探索长片段ROHs的现象及特征，从而拓展纯合子定位法范畴来鉴定数量性状位点，利用计算机模拟分析新方法的检测功效和假阳性率。通过在家畜数量性状和人类复杂疾病的全基因数据分析的应用，检验新方法的可靠性和适应性。最终将该方法做成软件，供研究人员使用。

数量性状位点的定位在复杂性状的遗传分析中占有重要地位。本研究基于全基因组“连锁不平衡”和“连锁”假设下建立一系列遗传统计方法来寻找、解释和检定纯合子片段与数量性状关系，主要包括有；①开发GWAS数据加性和显性效应的决策树统计学方法；②正向选择和有效群体数解释ROHs的现象及特征；③极值抽样设计鉴定ROHs与数量性状关系；④探索杂合子定位可行性。利用计算机模拟分析新方法的检测功效和假阳性率，检验新方法的可靠性和适应性。通过在家畜数量性状和人类复杂疾病的全基因数据分析，为人们了解复杂性状的遗传机理铺平道路。相应地，这一成果将有助于估计家畜 QTL 的育种值，可提高选种的准确性和育种效率；另外有助于预测人们罹患某些疾病的风险，并帮助他们在疾病的发生前积极预防。