Pangenome alignment reveals global diversity and evolution of human centromeric regions

Centromeres, crucial for chromosome segregation and genome stability, have long been understudied due to their repetitive and complex nature. Despite long-read sequencing and full genome assembly progress, aligning centromeres across populations remains difficult, hindering research into their mutation patterns. The Human Pangenome Reference Consortium recently sequenced more than 6,000 centromeres, offering the chance to map global centromere diversity.

However, centromeres were previously left out of pangenome alignments because of the technical difficulty of aligning their repetitive tandem repeats and highly variable structures. To tackle this issue, scientists have developed Centrolign, a graph-based multiple sequence alignment tool. Centrolign uses a uniqueness-driven objective function and partial-order alignment to precisely align alpha satellite higher-order repeats.

By focusing on rare matches within tandem arrays and utilizing extended centromere-spanning haplotypes resulting from suppressed recombination, Centrolign creates step-by-step alignments that maintain the original repeat organization. When applied to human centromeres, these alignments expose the evolutionary relationships between similar satellite array haplotypes and enable accurate calculations of variation rates, structural variant frequencies, and mutation patterns within satellite arrays.

By merging Centrolign-generated graphs with repeat annotation tools and pangenome mapping algorithms, researchers can accurately identify variants and genotype samples from long reads, even without prior assembly. Additionally, centromere haplotypes can be classified using k-mers alone. These advancements provide a reliable structure for integrating centromeres into larger pangenomes and population genomics studies, enhancing our knowledge of human genome evolution and diversity.