Evaluating relationship inference from low-quality DNA using a conditional simulation framework

Inference of genetic relationships from genomic data is central to applications in human, plant, and conservation genetics, with investigative genetic genealogy (IGG) increasingly used in forensic and population studies. While high-quality DNA samples enable robust detection of identity-by-descent (IBD) segments, low-quality or low-quantity sources such as telogen hairs pose challenges due to genotyping errors and incomplete genetic data (profiles). In this study, we 1) introduce a conditional simulation framework that generates relatives of varying degrees based on high-quality donor genotypes and allows systematic evaluation of kinship inference under controlled conditions, and 2) demonstrate that short fragments from single telogen hairs provide sufficient SNP data for IGG. Using buccal swabs and telogen hair samples from two donors, we generated whole-genome SNP profiles aided by imputations and compared them with simulated relatives. Despite the poor DNA yield from hair, imputation recovered > 78% of the 1,271,414 target IGG genotypes with > 99% concordance to reference profiles. Genotyping errors were primarily allelic dropouts, which were evenly distributed across the genome. Relationship inference revealed that hair-derived profiles generally supported accurate kinship classification, though elevated error rates in some samples reduced accuracy, particularly for close relatives. Adjusting IBD detection parameters, such as SNP set, error allowances and marker thresholds, improved classification without introducing false positives. In conclusion, our findings introduce a novel conditional simulation framework to quantify the impact of low quality, quantity, or coverage of DNA on kinship inference. We apply the framework to telogen hair samples and show the potential and limitations of single telogen hairs in IGG applications.