Göring H H, Ott J
Department of Genetics and Development, Columbia University, New York, N.Y. 10032, USA.
Eur J Hum Genet. 1997 Mar-Apr;5(2):69-77.
In linkage studies, errors in pedigree structure will often be uncovered through Mendelian inconsistencies. In affected sib pair analysis of diseases with late onset, however, such mistakes will usually go undetected since parental genotypes are commonly not known. Cases of nonpaternity, unrecorded adoption or accidental sample swap in the laboratory will then not be noticed. Typically, such relationship errors lead to a decrease in power for linkage. In this paper, a method is presented which allows verification of the relationship between stated sibs using their marker genotypes. The method is likelihood-based and incorporates a Bayesian approach to compute posterior relationship probabilities. It is shown that sibs, half-sibs and unrelated individuals can be distinguished from each other quite reliably using numbers of markers that should be available in most sib pair studies. It is demonstrated that elimination of false sib pairs increases the power to detect linkage in affected sib pair studies. The gain in power may be large if relationship errors occur quite frequently; the gain will be only moderate if relationship errors are very infrequent. Software for relationship estimation is provided.
在连锁研究中，系谱结构中的错误常常会通过孟德尔遗传不一致性被发现。然而，在对迟发性疾病进行受累同胞对分析时，由于通常不知道父母的基因型，这类错误往往难以被察觉。那么，实验室中出现的非亲生、未记录的收养或偶然的样本交换情况就不会被注意到。通常，这类关系错误会导致连锁分析的效能降低。本文提出了一种方法，该方法可以利用同胞的标记基因型来验证所宣称的同胞之间的关系。该方法基于似然性，并采用贝叶斯方法来计算后验关系概率。结果表明，使用大多数同胞对研究中应该能够获得的标记数量，可以相当可靠地区分同胞、半同胞和无血缘关系的个体。结果还表明，在受累同胞对研究中，消除错误的同胞对可以提高检测连锁的效能。如果关系错误频繁发生，效能的提高可能会很大；如果关系错误非常罕见，效能的提高则只会适中。本文还提供了关系估计软件。
