KinshipLR: software development and application for complex kinship testing in forensic genetics

Complex kinship testing presents significant challenges due to the complexity of genetic information among individuals and the lack of systematic exploration of genetic markers' effectiveness. This study developed an R package named KinshipLR to optimize biomarker selection for complex kinship testing, evaluating the efficacy of autosomal STRs, SNPs, and X-STRs across various scenarios. Using allele frequency data from Chinese four populations, 10,000 simulations were conducted for each scenario to analyze the factors influencing testing accuracy, including relationship type, population background, genetic marker type, and evaluation algorithms. The findings highlight that the efficacy of genetic markers in complex kinship testing varies among populations. Moreover, while the standalone performance of 20 CODIS-STRs is limited, adding 27 X-STRs significantly improves accuracy in testing scenarios such as full siblings, grandmother-granddaughter, aunt-niece, and half-siblings. Additionally, integrating the IBS with SVM method outperforms traditional LR-based threshold method in complex kinship testing. This study provides a tailored approach for genetic marker selection, enhancing the efficiency and reliability of complex kinship testing and offering valuable insight for forensic application.