Abstract
<title>Abstract</title> <p>Short tandem repeats (STRs) are abundant components of the genome that contribute to genetic variation and genome instability and have been implicated in numerous human diseases, particularly neurological disorders. Although STR instability has been extensively characterized at individual loci, the broader evolutionary forces shaping STR sequence distributions across the genome remain incompletely understood. Here, we performed a genome-wide computational analysis of STR motifs in the human genome, integrating comparative primate genomics with cell-type-specific expression data to investigate sequence depletion, evolutionary conservation, and potential functional associations. We identified two distinct classes of STR sequence behavior. First, CpG-containing motifs showed pronounced depletion relative to genomic expectation, with several motifs reduced by more than 26,000-fold. These patterns are consistent with methylation-associated mutational processes and long-term sequence erosion. Second, among motifs with available comparative conservation data, we identified a subset of 11 non-CG motifs displaying elevated conservation across primate evolution, suggesting the possibility of long-term evolutionary constraint associated with intrinsic sequence properties, including repetitive symmetry and other intrinsic sequence features. Notably, depletion within the human genome was not significantly correlated with evolutionary conservation across the analyzed motif subset, indicating that these signatures likely reflect partially independent evolutionary processes. To evaluate potential functional associations, we compared expression profiles of genes located near conserved non-CG motifs with distance-matched control genes across six human cell types using CZ CELLxGENE datasets. The final analysis of 10 target genes proximal to conserved motifs showed significantly higher expression specifically in pancreatic beta cells (approximately 1.76-fold higher mean scaled expression, p = 0.00118), whereas no significant differences were observed in the other analyzed cell types. We further investigated whether this expression enrichment is altered in diabetic states. In type 1 diabetes, the expression advantage was lost (fold change = 1.11, p = 1.00), whereas in type 2 diabetes, the pattern reversed, with target genes showing lower expression than controls (fold change = 0.84, p = 1.00). DNA methylation analysis revealed that these changes were not accompanied by methylation alterations in type 1 diabetes; however, in type 2 diabetes, four conserved motifs (AAGA, GAGG, TATG, ATCT) showed significant hypomethylation (fold change = 0.53, p = 0.0034). These findings suggest that conserved non-CG motifs function as condition-dependent regulatory elements, with distinct molecular mechanisms underlying beta-cell dysfunction in type 1 and type 2 diabetes. In addition, chromosome X displayed STR architectures characterized by shorter repeat lengths, smaller repeat periods, and lower complexity scores relative to autosomes, consistent with long-term influences of chromatin context and epigenetic state on repeat organization. Together, these findings support a model in which epigenetic mutational bias, long-term structural constraint, and genomic context independently contribute to shaping the repetitive sequence landscape of the human genome.</p>