Duncan G T, Noppinger K, Carey J, Tracey M
Department of Biological Sciences, Florida International University, Miami 33199.
Genetica. 1993;88(1):51-7. doi: 10.1007/BF02424451.
There is considerable debate about the methodologies used to estimate VNTR (Variable Number of Tandem Repeats) multi-locus genotype frequencies or odds of inclusion in forensic cases. To compare two of the methods in use, allele frequency distributions among six populations were compared and the effect of population heterogeneity on VNTR multi-locus genotype frequency estimation was examined. Genotype frequencies estimated from single population data were one or two orders of magnitude smaller than those estimated by picking the highest allele frequency in a group of subpopulations to estimate genotype frequencies using a ceiling principle. The average change does not appear to be very sensitive to the set of subpopulations used; four locus frequencies still give inclusion odds of one in a million or less. We think that use of the ceiling principle solves both the statistical problem engendered by subpopulation heterogeneity and the legal problem of assuming that the prepetrator and suspect belong to the same subpopulation. The counterintuitive fact of human genetic polymorphism is that it is easier to identify an individual than it is to identify the subpopulation, ethnic group or race to which that individual belongs.
关于用于估计可变数目串联重复序列（VNTR）多位点基因型频率或法医案件中包含几率的方法，存在相当多的争论。为了比较两种常用方法，我们比较了六个群体中的等位基因频率分布，并研究了群体异质性对VNTR多位点基因型频率估计的影响。从单群体数据估计的基因型频率比通过在一组亚群体中挑选最高等位基因频率并使用上限原则估计基因型频率所得的值小一到两个数量级。平均变化似乎对所使用的亚群体集合不太敏感；四个位点频率的包含几率仍然在百万分之一或更低。我们认为，使用上限原则既解决了亚群体异质性带来的统计问题，也解决了假设犯罪者和嫌疑人属于同一亚群体的法律问题。人类遗传多态性的一个违反直觉的事实是，识别一个个体比识别该个体所属的亚群体、族群或种族更容易。
