Systematic Evaluation of Variants Linked to Hearing Loss using Minor Allele Frequency and Prediction Tools

Abstract

Non-syndromic hearing loss (NSHL) is extremely genetically heterogeneous, and to date, more than 96 genes have been linked to NSHL and explain about half of the clinical cases. Although high throughput DNA sequencing technology facilitates the identification of causative mutations in many human diseases, hundreds or thousands of variants identified by this method require interpretation to assess their likelihood of causing a disease. Here, we aim to systemically evaluate variants in 96 genes, which have been identified in NSHL patients, using minor allele frequency (MAF) and predictive tools. The MAF thresholds were determined considering allele frequency of the most common pathogenic variant of GJB2, and the prevalence of NSHL. For the 96 NSHL known genes, 3,082 variants reported in HGMD and 1,210 reported as pathogenic or likely pathogenic in ClinVar were classified according to the MAF threshold and then according to the pLI scores of corresponding genes into three categories (pLI<0.1, 0.1<pLI<0.9, pLI>0.9). The number of missense variants reported in recessive (rec), dominant (dom) and dom/rec genes was 1,040, 244, and 668 respectively. The prediction scores of the missense variants were obtained using PolyPhen-2, SIFT, and Condel. As a result of analysis, the variants above the MAF threshold were 61, 23 and 14 in recessive, dominant and dom/rec genes, respectively. Using Korean control dataset, three variants that would be found more frequently in Koreans than in any other population were identified suggesting that several variants having MAF levels which are implausible for highly penetrance Mendelian disease could be found through other certain population control datasets. Additionally, there were statistical differences in prediction scores between the variants below and above the MAF threshold in recessive genes. Although prediction scores were not different between the variants below and above the MAF threshold for dominant genes, the scores were significantly different for dominant genes with > 0.9 pLI score. These data showed that prediction tools could be more useful for predicting variants in recessive genes and dominant genes with > 0.9 pLI score. Based on this study, we can prioritize novel candidate variants that have a causal relationship with the disease by using the MAF threshold and the prediction tool to evaluate variants in NSHL.