Answers are grounded in this paper and cite their sources.
Short tandem repeats are among the most error-prone sequences in short-read genome catalogs, and several sit inside loci that clinical screening depends on. One such locus is the cystic fibrosis transmembrane conductance regulator (CFTR) (TG)mTn microsatellite (the polypyrimidine tract of intron 9), whose 5T allele—a five-thymine splice-acceptor variant that increases skipping of exon 10 (legacy exon 9)—is a standard reflex target when the R117H (a CFTR missense variant whose pathogenicity depends on the 5T configuration in cis) is detected and in CBAVD/CFTR-related-disorder work-ups. Re-typing this tract at read level in 100 genotype-stratified 1,000 Genomes donors, we show that the high-coverage 30 × 1,000 Genomes reference call set reports no classic poly-T deletion (TG11-T5) allele in any of the 3,202 donors, yet read-level re-typing recovers nine genuine 5T chromosomes and resolves all ten tract configurations that the catalog could not decompose. This clinically used allele—segregating at up to ∼5.4% depending on ancestry (AFR 5.4%, EUR 3.0%, global 2.3%)—is, thus, entirely invisible to the flagship reference catalog. We reconstructed four-locus (TG)m–Tn–M470V–rs11974978 haplotypes for 6,404 chromosomes across five super-populations, finding that rs213950 (M470V) is frequency-stable outside Africa but tags markedly different haplotype backgrounds by linkage. Exon 10 (legacy exon 9) skipping, measured from public RNA-seq junction counts, was ∼9.5% in the pancreas and ∼15% in the colon, ileum, and salivary gland (pooled junction-count estimates) and was higher in the colon than in the pancreas within the same 94 donors (paired Wilcoxon P = 8.9 × 10 −5 ). The intron-10 SNP rs11974978 is a splicing quantitative trait locus QTL (sQTL) for exon 10 skipping, found in the GTEx pancreas catalog and independently corroborated as the lead splicing single-nucleotide polymorphism (SNP) in 399 pancreatic islets (P = 3.6 × 10 −21 ). A cis-additive four-locus model predicts that East Asians carry the highest population skipping background (mean 26%) and Africans the lowest (15%); a random 300-donor re-typing confirms the rankings are robust to tract-label errors. Read-level interrogation or targeted genotyping, rather than variant-call format (VCF) calls, is required at this clinically used tract. More broadly, these results show that a clinically used microsatellite allele can be entirely absent from a flagship high-coverage short-read reference catalog and that read-level or targeted re-genotyping—resolved as full haplotypes, not single markers—is required before population catalogs can inform splicing-trait interpretation or carrier screening.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: