Testing of two SNP array–based genealogy algorithms using extended Han Chinese pedigrees and recommendations for improved performances in forensic practice

• Published in: Electrophoresis Vol. 44; no. 17-18; pp. 1435 - 1445
• Main Authors: Liu, Jing, Wei, Yi‐Liang, Yang, Lan, Jiang, Li, Zhao, Wen‐Ting, Li, Cai‐Xia
• Format: Journal Article
• Language: English
• Published: Germany Wiley Subscription Services, Inc 01.09.2023
• Subjects: Algorithms; ancestry; Chinese people; crime; data collection; Datasets; electrophoresis; forensic sciences; Genealogy; germ cells; Inference; kinship; meiosis; progeny; Segments; statistics; identity-by-state; long-range familial search; single nucleotide polymorphism; kinship analysis; identity-by-descent
• ISSN: 0173-0835; 1522-2683; 1522-2683
• DOI: 10.1002/elps.202200237

Distant genetic relatives can be linked to a crime scene sample by computing identity‐by‐state (IBS) and identity‐by‐descent (IBD) shared by individuals. To test the methods of genetic genealogy estimation and optimal the parameters for forensic investigation, a family‐based genetic genealogy analysis was performed using a dataset of 262 Han Chinese individuals from 11 families. The dataset covered relative pairs from 1st‐ to 14th degrees. But the 7th‐degree relative is the most distant kinship to be fully investigated, and each individual has ∼200 relatives within the 7th degree. The KING algorithm by calculating IBS and IBD statistics can correctly discriminate the first‐degree relationships of monozygotic twin, parent–offspring and full sibling. The inferred relationship was reliable within the fifth‐degree, false positive rate <1.8%. The IBD segment algorithm, GERMLINE + ERSA, could provide reliable inference result prolonged to eighth degree. Analysis of IBD segments produced obviously false negative estimations (<27.4%) rather than false positives (0%) within the eighth‐degree inferences. We studied different minimum IBD segment threshold settings (changed from >0 to 6 cM); the inferred results did not make much difference. In distant relative analysis, genetically undetectable relationships begin to occur from the sixth degree (second cousin once removed), which means the offspring after seven meiotic divisions may share no ancestor IBD segment at all. Application of KING and GERMLINE + ERSA worked complementarily to ensure accurate inference from first degree to eighth degree. Using simulated low call rate data, the KING algorithm shows better tolerance to marker decrease compared with the GERMLINE + ERSA segment algorithm.