GPU-accelerated linkage disequilibrium scans reveal non-independent assortment of human non-homologous chromosomes
摘要
The independence of assortment between loci on non‑homologous chromosomes, as proposed by Mendel’s laws, has not been rigorously validated at the whole‑genome scale due to the computational demands of estimating inter‑chromosomal linkage disequilibrium.
ResultsUsing a GPU-accelerated method, we calculate approximately 100 trillion linkage disequilibrium coefficients among 22 non-homologous chromosomes across 26 populations from the 1000 Genomes Project. Contrary to the assumption of independent segregation of non-homologous chromosomes during meiosis, linkage disequilibrium mapping of 2,594 samples reveals 1,195 locus pairs with high linkage disequilibrium (r2 ≥ 0.8), clustered near centromere regions. This pattern is consistently observed across super‑populations and individual populations. Seven populations of African ancestry have the fewest high linkage disequilibrium inter-chromosomal pairs, while five populations of East Asian ancestry show the most high linkage disequilibrium pairs, mirroring known population evolutionary history.
ConclusionsThese findings indicate that human non‑homologous chromosomes do not assort fully independently in the centromere region, suggesting possible inter‑centromeric haplotypes.