Yi N, Xu S
Department of Botany and Plant Sciences, University of California, Riverside, California 92521, USA.
Genetics. 2001 Apr;157(4):1759-71. doi: 10.1093/genetics/157.4.1759.
Quantitative trait loci (QTL) are easily studied in a biallelic system. Such a system requires the cross of two inbred lines presumably fixed for alternative alleles of the QTL. However, development of inbred lines can be time consuming and cost ineffective for species with long generation intervals and severe inbreeding depression. In addition, restriction of the investigation to a biallelic system can sometimes be misleading because many potentially important allelic interactions do not have a chance to express and thus fail to be detected. A complicated mating design involving multiple alleles mimics the actual breeding system. However, it is difficult to develop the statistical model and algorithm using the classical maximum-likelihood method. In this study, we investigate the application of a Bayesian method implemented via the Markov chain Monte Carlo (MCMC) algorithm to QTL mapping under arbitrarily complicated mating designs. We develop the method under a mixed-model framework where the genetic values of founder alleles are treated as random and the nongenetic effects are treated as fixed. With the MCMC algorithm, we first draw the gene flows from the founders to the descendants for each QTL and then draw samples of the genetic parameters. Finally, we are able to simultaneously infer the posterior distribution of the number, the additive and dominance variances, and the chromosomal locations of all identified QTL.
数量性状基因座（QTL）在双等位基因系统中易于研究。这样的系统需要两个近交系杂交，这两个近交系可能对于QTL的替代等位基因是固定的。然而，对于世代间隔长且存在严重近交衰退的物种，近交系的培育可能既耗时又成本高昂。此外，将研究限制在双等位基因系统有时可能会产生误导，因为许多潜在重要的等位基因相互作用没有机会表达，因而无法被检测到。涉及多个等位基因的复杂交配设计模拟了实际的育种系统。然而，使用经典的最大似然法很难开发统计模型和算法。在本研究中，我们研究了通过马尔可夫链蒙特卡罗（MCMC）算法实现的贝叶斯方法在任意复杂交配设计下进行QTL定位的应用。我们在一个混合模型框架下开发该方法，其中将奠基者等位基因的遗传值视为随机的，而非遗传效应视为固定的。利用MCMC算法，我们首先为每个QTL绘制从奠基者到后代的基因流，然后绘制遗传参数的样本。最后，我们能够同时推断所有已鉴定QTL的数量、加性和显性方差以及染色体位置的后验分布。
