Institute of Medical Informatics and Statistics, Christian-Albrechts University, Campus Kiel, House 31, Arnold-Heller-Str. 3, 24105, Kiel, Germany.
Int J Legal Med. 2010 May;124(3):205-15. doi: 10.1007/s00414-009-0413-0. Epub 2010 Feb 10.
At least in principle, most instances of complex kinship testing can be reduced to pairwise kinship cases involving two critical family members that either link or separate presumed sub-branches of a family. In the European population, the 34 short tandem repeats (STRs) currently used in forensic genetics are sufficiently powerful to allow assessment of disputed first and second but not lower degrees of pairwise blood relatedness. We provide estimates of the means and variances of marker-specific log-likelihood ratios, using large-sample approximation and assuming different scenarios of pairwise kinship analysis. These estimates allow power calculations to be performed for any combination of the available STRs. Since some of the markers considered are physically linked, chromosome-wide likelihood calculations in kinship cases other than parent-child duos (and trios) have to take the reduced rates of meiotic inter-marker recombination into account. We show by simulation that this requirement may be ignored when discriminating distant hypotheses about kinship, but that linkage may play an important role in the biostatistical analysis of more intricate cases.
原则上，大多数复杂亲属关系测试的实例都可以简化为涉及两个关键家庭成员的成对亲属关系案例，这两个关键家庭成员要么连接，要么分离家庭的假定分支。在欧洲人群中，目前在法医学遗传学中使用的 34 个短串联重复序列（STR）足以评估有争议的第一和第二但不是更低程度的成对血缘关系。我们使用大样本逼近并假设成对亲属关系分析的不同情况，提供了标记特异性对数似然比的均值和方差的估计。这些估计允许针对可用 STR 的任何组合进行功效计算。由于所考虑的一些标记物是物理上相连的，因此除了亲子二人组（和三人组）之外的亲属关系案例中的全染色体似然计算必须考虑减数分裂过程中标记物间重组的降低速率。我们通过模拟表明，在区分关于亲属关系的遥远假设时，可以忽略这一要求，但在更复杂的情况下，连锁可能在生物统计学分析中发挥重要作用。
