Calculus Institute, University of Buenos Aires, Argentina; IDEPI, National University of Jose Clemente Paz, Argentina.
Calculus Institute, University of Buenos Aires, Argentina.
Forensic Sci Int Genet. 2023 Sep;66:102891. doi: 10.1016/j.fsigen.2023.102891. Epub 2023 Jun 9.
The search for missing persons implies several steps, from the preliminary investigation that involves collecting background data related to the case to the genetic kinship testing. Despite its crucial importance in identifications, only some approaches mathematically formalize the possibility of using preliminary investigation data. In some cases, a filtering strategy is applied, which implies selecting a subset of possible victims where some non-genetic variables perfectly match those of the missing. Through a Bayesian approach, we propose a mathematical model for computing the prior odds based on non-genetic variables usually collected during the preliminary investigation, such as biological sex, hair colour, and age. We use computational simulations to show how to incorporate these prior odds in DNA-database searches. Importantly, our results suggest that applying the proposed model leads to better search performance in underpowered cases from the genetic point of view, where few or distant relatives of the missing person are available for genotyping. Furthermore, the results are also helpful when using non-genetic data for prior odds in well-powered cases, where genetic data are enough to reach a reliable conclusion. It performs better than other approaches, such as using non-genetic data for filtering. The software mispitools, freely available on CRAN, implements all described methods (https://CRAN.R-project.org/package=mispitools).
寻找失踪人员需要几个步骤，从初步调查开始，其中包括收集与案件相关的背景数据，到基因亲属关系测试。尽管它在识别方面至关重要，但只有一些方法从数学上形式化了使用初步调查数据的可能性。在某些情况下，应用过滤策略，即选择一些可能的受害者子集，其中一些非遗传变量与失踪者的完全匹配。通过贝叶斯方法，我们提出了一种基于通常在初步调查中收集的非遗传变量（如生物性别、头发颜色和年龄）计算先验几率的数学模型。我们使用计算模拟来展示如何将这些先验几率纳入 DNA 数据库搜索中。重要的是，我们的结果表明，在遗传角度上，在能力不足的情况下应用所提出的模型可以提高搜索性能，在这种情况下，失踪者的亲属人数较少或较远，无法进行基因分型。此外，当在能力较强的情况下使用非遗传数据进行先验几率时，结果也很有帮助，此时遗传数据足以得出可靠的结论。它比其他方法表现更好，例如使用非遗传数据进行过滤。mispitools 软件可在 CRAN 上免费获得，它实现了所有描述的方法（https://CRAN.R-project.org/package=mispitools）。
