The coalescent process in genetics : mapping complex diseases and other applications

聚结过程是一种用来模拟群体谱系的随机过程，在过去的15年里引起了人们的广泛关注，主要是由于它在人类遗传学研究中的大量应用。我们已经介绍了一种使用具有重组的聚结过程的方法（Larribe等人，2002，Larribe，2004），以便估计影响疾病的基因突变的位置，即以便进行基因精细作图。在遗传标记的“窗口”上使用复合条件似然策略（Larribe和Lessard，2008），我们提高了估计的稳定性，并大大增加了该方法可用的遗传标记的数量。然后，我们通过设计用于从表型和基因型估计单倍型和致病等位基因的方法，在复杂遗传疾病的背景下进一步发展了该方法（Larribe et al.，2015年）;此外，该方法现在能够考虑允许不完全复制和表型复制的遗传模型。遗传学领域的常见做法仍然是基于数千个关联统计数据进行分析，每个关联统计数据针对一对遗传标记，从而忽略个体之间的依赖性和标记之间的空间遗传依赖性。因此，在我们的建议中，我们继续完成一项艰巨的任务，即通过解决上述两个依赖性问题，为这一问题提供一个完善的解决方案。与许多其他方法相比，我们的方法不是基于测试关联，而是基于可能性的遗传作图方法;因此，它结合了丰富的遗传模型。

The coalescent process, a stochastic process used to simulate population genealogies, has attracted much attention during then past fifteen years, mostly due to the numerous applications in human genetics research. We have introduced a method that uses the coalescent process with recombination (Larribe et al., 2002, Larribe, 2004), in order to estimate the location of a genetic mutation influencing a disease, i.e. in order to do genetic fine mapping. Using a composite-conditional-likelihood strategy (Larribe and Lessard, 2008) on "windows" of genetic markers, we have improved the stability of the estimates and greatly increased the number of genetic markers usable by the method. Then, we have further developed the approach in the context of complex genetic diseases, by devising methods for estimating haplotypes and causal alleles from phenotypes and genotypes (Larribe et al., 2015); moreover, the method is now able to take into account genetic models allowing incomplete penetrance and phenocopy. It is still common practice in the genetics world to base the analysis on thousands of association statistics, each for a pair of genetic markers, and thus to ignore the dependence between individuals and the spatial genetic dependency between markers. Therefore, in our proposal we continue the difficult task of offering a refined solution to this problem by addressing both dependency issues mentioned above. In contrast with many other approaches, our method is not based on testing association, but is a genetic mapping methodology based on the likelihood; as such, it incorporates rich genetic models.