Gilbert's syndrome is mainly diagnosed through genetic analysis and is primarily detected through a mutation in the promoter region of the UGT1A1 gene. However, most of the research has been conducted on Caucasian populations. In this study, we studied the Han population in Taiwan to investigate the possibility of other mutations that could cause Gilbert's syndrome. This study comprised a test group of 45 Taiwanese individuals with Gilbert's syndrome and 180 healthy Taiwanese individuals as a control group. We extracted DNA from the blood samples and then used Axiom Genome-Wide TWB 2.0 array plates for genotyping. Out of 302,771 single nucleotide polymorphisms (SNPs) from 225 subjects, we detected 57 SNPs with the most significant shift in allele frequency; 27 SNPs among them were located in the UGT1A region. Most of the detected SNPs highly correlated with each other and are located near the first exon of UGT1A1, UGT1A3, UGT1A6, and UGT1A7. We used these SNPs as an input for the machine learning algorithms and developed prediction models. Our study reveals a good association between the 27 SNPs detected and Gilbert's syndrome. Hence, this study provides a reference for diagnosing Gilbert's syndrome in the Taiwanese population in the future.
Keywords: Gilbert’s syndrome; genetic factors; machine learning; single nucleotide polymorphism.