AUCP: An indicator for system effectiveness of panels in pairwise distant kinship identification

Abstract

Multiple studies have been done for the identification of pairwise distant kinship and several targeted panels have been constructed. For most of such constructions, pedigree analysis was applied to evaluate the system effectiveness of a certain panel. However, such analyses were hard to be compared to each other and could be affected by many factors, such as sample size and sampling randomness. A new indicator named predicted area under ROC curve (AUCP), where ROC curve stood for receiver operating characteristic curve, was derived applying binomial distribution theory and analyzed with simulated and real cases in this study. After comparing between the values of AUCPs and results of pedigree analyses with different loci sets and kinship types, the ability of these two methods evaluating the system effectiveness was proved to be close to each other. The implementation of AUCP was much easier than pedigree analysis, because a secondary sampling or simulation was not needed. Therefore, AUCP can be a better indicator for panels targeted to pairwise distant kinship identification and we are recommending it as an indicator calculated by default for such panels.