Construction and validation of a preimplantation kinship identification model using ultra-low-depth whole genome sequencing

Errors during in vitro fertilization (IVF) workflows pose significant risks; yet, existing quality control measures remain insufficient. Herein, we developed a preimplantation full-sibling relationship identification (PFSRI) model using ultra-low-depth whole genome sequencing (uldWGS) data from 789 biopsied embryos (average depth, 0.02×). The model predicts embryonic relationships via the heterozygosity-based relationship ratio (HetRR). In double-blind testing, PFSRI successfully distinguished full-sibling embryos from different ovarian stimulation cycles within the same family and identified resequenced samples. At optimized thresholds (unique reads [URs] ≥ 800 K and co-detected SNPs [CoSNPs] ≥ 800), the model achieved 100% accuracy in discriminating full-sibling, unrelated, and identical relationships, with performance unaffected by chromosomal status. The PFSRI model provides a cost-effective and robust tool for precise embryo relationship identification, serving as a valuable quality control instrument for IVF laboratories.