Pangenome alignment reveals global diversity and evolution of human centromeric regions

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Centromeres play essential roles in chromosome segregation and genome stability, yet they remain among the least characterized regions of the human genome. Despite advances in long-read sequencing and complete genome assembly, the extreme repetitiveness and structural complexity of these regions still challenge population-scale analysis, obscuring their mutational dynamics. The Human Pangenome Reference Consortium has now accurately assembled over 6,000 centromeres, providing an opportunity to catalog global centromere variation. However, centromeric regions have been systematically excluded from pangenome alignments due to the technical challenge of aligning their highly repetitive tandem arrays and extreme structural variability. Here we introduce Centrolign, a graph-based multiple sequence alignment tool that combines a uniqueness-driven objective function with partial-order partial-order alignment to accurately align alpha satellite higher-order repeats. By prioritizing rare matches within tandem arrays and leveraging extended centromere-spanning haplotypes formed by suppressed recombination, Centrolign produces progressive multiple sequence alignments that preserve ancestral repeat organization. Applied across human centromeres, these alignments reveal the phylogenetic structure of similar satellite array haplotypes and enable precise estimation of variation rates, structural variant frequencies, and spatial patterns of mutation within satellite arrays. Integrating Centrolign graphs with repeat annotation tools and pangenome mapping algorithms allows accurate variant calling and genotyping from long reads without prior assembly. Moreover, we show that centromere haplotypes can be accurately subtyped with k-mers alone. Together, these advances establish a robust framework for incorporating centromeres into broader pangenomes, and population genomics in general, advancing our understanding of human genome evolution and diversity.
Funding Information
  • NIH/NHGRI (R01HG011274)
  • NIH/NHGRI (UM1HG010971)
  • NIH/NHGRI (R01HG014490)
  • NIH/NHGRI (U01HG010961)
  • NIH/NHGRI (U24HG010262)
  • NIH/NHGRI (OT2OD026682)

This publication has 82 references indexed in Scilit: