Influence of population structure on X‑chromosome SNP‑based inbreeding coefficients and sex‑verification thresholds in pig breeding populations
Abstract
Background: This study evaluated how population structure affects a non‑pseudoautosomal X chromosome score (F X ) used to verify recorded sex in pig breeding populations. Analyses included 10,368 Duroc, Landrace, and Yorkshire pigs and 2,385 single nucleotide polymorphisms (SNPs) selected using recorded females. Biological sex was not independently verified; performance therefore denotes agreement with breeding records. Results: On the pooled female reference, mean female F X was 0.394, 0.145, and 0.223 in the 3 breeds. Holding all 2,385 SNPs fixed and recalculating only within‑breed allele frequencies reduced these means to 0.013, 0.052, and 0.030 and their range by 84.3%. The all‑animal area under the receiver operating characteristic curve (AUC) was 0.997, with 99.61% agreement at a cutoff of 0.883 and 40 discordant animals. In 10 repetitions of sire‑grouped 5‑fold cross‑validation, marker selection, reference estimation, and cutoff selection were restricted to training animals; mean AUC was 0.9967 and mean agreement was 99.59% (range, 99.58–99.60%). Ten repetitions of dam‑grouped validation and leave‑one‑farm‑out validation yielded mean agreement of 99.59% (range, 99.57–99.60%) and agreement of 99.60%, respectively. Breed‑specific cutoffs did not consistently improve validation agreement. The causes of discordance remain unresolved. Conclusion: These findings support population‑specific calibration and validation that accounts for the grouping of animals in breeding populations. They do not establish a universal cutoff or accuracy against independently confirmed biological sex.