Human centromeres are thought to be highly diverse and rapidly evolving but have been difficult to sequence owing to their repetitive nature, consisting of tandemly repeating α-satellite DNA organized into higher-order repeat (HOR) units. Logsdon et al. report the second complete sequence of a set of human centromeres, enabling comparative analyses of sequence and structural variation.

Three alignment strategies were used to compare CHM1 and CHM13 centromeres, showing that 63.0–71.5% of α-satellite HORs (depending on the chromosome) could be reliably aligned. This dropped to 53.2–55.3% for alignment to centromeres from 56 HPRC and HGSVC genomes, which illustrates the considerable sequence variation between human haplotypes. The sequence identity of CHM1 and CHM13 centromeres generally increased with increasing distance from the core of the α-satellite HORs into the monomeric α-satellite DNA in the pericentromeres. Overall, the analyses show a high level of sequence diversity in the α-satellite HORs of human centromeres, but sequence divergence varies considerably depending on the chromosome; chromosomes 19 and X had the greatest sequence identity, whereas chromosome 5 had as much as 4% sequence divergence. This chromosome-specific variance in sequence divergence was confirmed by comparison with the HPRC and HGSVC centromeres.