Statistical and Bioinformatics Methods

Hardy-Weinberg equilibrium was tested for each SNP within the control samples. χ² tests were performed for each SNP to evaluate the differences in allelic and genotypic distributions between UL cases and controls. Linkage disequilibrium (LD) blocks were constructed for both genes, and haplotype-based analyses were conducted for each block. Plink was utilized for the analyses mentioned above¹⁸. In addition to genetic association analyses focusing on disease status, we also analysed the potential link between significant SNPs and four clinical features of UL, including bleeding, pain, number of fibroid nodes, and size of the node, in a subset of our samples that included UL cases only. χ² tests were performed for these analyses. In general, Bonferroni correction was applied to address multiple comparisons. For single marker-based association analyses, the threshold P value was 0.05/55 ≈ 9 × 10⁻⁴. Genomic control was applied to correct for the potential effects of population stratification^19,20. The null distribution of genomic inflation factor λ was constructed by 10,000 bootstrapping.

The potential biological functions of our selected SNPs were evaluated through RegulomeDB (http://www.regulomedb.org/)²¹. RegulomeDB is a database that annotates SNPs based on known and predicted regulatory element data from the ENCODE project. A score ranging from 1–6 was assigned to each SNP, and a lower score indicated a more significant biological function. In addition, we also extracted eQTL data from the GTEx database (https://www.gtexportal.org/home/)²² to examine differences in gene expression associated with our significant SNPs.