Suzanne de Bruijn

292 Chapter 6 including deep-learning and machine-learning tools, have been developed to aid in the interpretation of candidate variants. An example is the SpliceAI deep-learning tool designed to assess putative splice variants. 47 This deep-learning tool shows significant advances in in silico splice effect predictions, as it does not depend on preselected features. 48 There is a growing interest to implement the deep-learning strategy to assess regulatory variants and SVs as well. 49,50 Additionally, machine learning tools are also employed for predictive and diagnostic purposes. Audio profiling software has been reported that can accurately predict, based on audiometric data, whether a person is a carrier of pathogenic variants associated with specific types of HL. 51,52 Artificial intelligence also offers possibilities to investigate oligogenic or multifactorial diseases. Machine learning tools to predict potential pathogenicity of a combination of oligogenic variants in an individual have been described (e.g. ORVAL 53 and Variant Combinations Pathogenicity Predictor 54 ). Using a machine-learning approach such as Variant Combinations Pathogenicity Predictor, it is possible to accurately predict potential dialellic inheritance. With the increasing sizes of available sequence datasets and computational power, statistical testing has become possible. Statistical tools can be used to e.g. investigate gene-specific enrichment of de novo pathogenic variants 55 , or genome-wide association studies (GWAS) studies can be performed to identify genetic risk loci for multifactorial disease. Several GWAS studies have been reported to investigate potential genetic risk factors underlying age-related HL. Whereas initially no loci could be identified with genome-wide significance 56 , by investigating a larger patient cohort, 44 independently associated genomic loci were found. 57 While still in an early stage, it can be speculated that computational and artificial intelligence tools will increasingly contribute to variant identification and interpretation in the following years. WHICH CHALLENGES ARE STILL AHEAD? Last year, the first complete gapless human chromosome sequence was obtained. The X chromosome was successfully sequenced from telomere-to-telomere, which was achieved by combining high-coverage ultra-long nanopore sequencing (Oxford Nanopore). 58 Scientists were finally able to resolve the remaining 29 gaps of the X chromosome that were still present in the reference genome, which mostly consisted of repeat-rich sequences, and herewith completion of the human genome has become within reach. This achievement is considered an important step forward towards the