The genetic architecture of pentanucleotide repeat variation and disease association.

Pentanucleotide tandem repeats (PNTRs) are associated with at least eleven neurological disorders in which pathogenic alleles are defined by both large repeat expansions and disease-specific non-reference motifs. However, population-scale, sequence-resolved datasets of PNTR variation have not been available, limiting interpretation of repeat alleles identified through long-read sequencing. Here, we genotyped 47,952 PNTR loci in 543 high-coverage PacBio HiFi genomes spanning five ancestry superpopulations, with validation in 500 Oxford Nanopore genomes. Using base-pair-resolved decomposition, we generated a genome-wide reference of PNTR length and sequence variation. Disease-associated PNTR loci emerged as extreme outliers for repeat-length variability, extreme allele length, and motif diversity. Sequence-resolved analyses further revealed that repeat length and sequence composition are frequently coupled, whereby alleles of distinct motif and interruption configurations occupy discrete length ranges. Finally, haplotype-resolved analyses showed that repeat variability and disease association converge on specific chromosomal backgrounds, exemplified by the SAMD12 locus. Together, these findings establish principles of PNTR variation for interpreting repeat alleles, prioritizing candidate loci, and discovering novel repeat expansion disorders.