Detecting large genomic differences, known as structural variants (SVs), remains one of the most persistent hurdles in genetics research. A new preprint by Qian Qin and Heng Li examines why these variants are especially difficult to identify within repetitive stretches of DNA called low-complexity regions (LCRs).
Although LCRs make up just 1.2 percent of the human reference genome, the authors found that they contain nearly 70 percent of validated SVs in the well-studied HG002 benchmark sample. The study compared multiple long-read SV calling tools and reported that between 77 and 91 percent of erroneous calls occur within LCRs. The frequency of mistakes increased with the length of the repeated sequences, underscoring how problematic these regions are for existing algorithms.
The researchers traced many of the errors to inconsistent alignment of sequencing reads in long repeats, which confuses variant callers. Methods that incorporated haplotype-aware realignment performed better but still faced challenges with large repeats. Even with high-quality long-read data, some callers missed up to half of the variants in LCRs over two kilobases.
The work suggests that SVs in LCRs should be considered a distinct class when benchmarking or analyzing genetic variation. The authors note that while some LCRs overlap with functional genes and cannot be ignored, careful stratification and algorithmic improvements are essential for accurate analysis. They also emphasize the promise of haplotype-resolved assemblies and pangenome-based approaches, which may provide more reliable variant calls in these complex regions.
Publication Details
Qin Q, Li H. Challenges in structural variant calling in low-complexity regions. arXiv. Preprint published September 27, 2025. doi:10.48550/arXiv.2509.23057
Although LCRs make up just 1.2 percent of the human reference genome, the authors found that they contain nearly 70 percent of validated SVs in the well-studied HG002 benchmark sample. The study compared multiple long-read SV calling tools and reported that between 77 and 91 percent of erroneous calls occur within LCRs. The frequency of mistakes increased with the length of the repeated sequences, underscoring how problematic these regions are for existing algorithms.
The researchers traced many of the errors to inconsistent alignment of sequencing reads in long repeats, which confuses variant callers. Methods that incorporated haplotype-aware realignment performed better but still faced challenges with large repeats. Even with high-quality long-read data, some callers missed up to half of the variants in LCRs over two kilobases.
The work suggests that SVs in LCRs should be considered a distinct class when benchmarking or analyzing genetic variation. The authors note that while some LCRs overlap with functional genes and cannot be ignored, careful stratification and algorithmic improvements are essential for accurate analysis. They also emphasize the promise of haplotype-resolved assemblies and pangenome-based approaches, which may provide more reliable variant calls in these complex regions.
Publication Details
Qin Q, Li H. Challenges in structural variant calling in low-complexity regions. arXiv. Preprint published September 27, 2025. doi:10.48550/arXiv.2509.23057
