An Established Relationship

A new approach to creating a genomics database could move the field one step closer to curating the clinical exome for gene–disease relationships. The approach combines a human element – systematic literature review – with AI indexing of millions of published abstracts and full-text references with genetic information. Results of a study testing the new approach, presented at the Annual Clinical Genetics Meeting 2024, identified over 10,000 germline gene–disease relationships – more than half of which showed positive associations with a specific disease (1).

Mark Kiel, Chief Scientific Officer of Genomenon – and one of the study leads – explains the drivers for the project: “It is estimated that about one-third of all genomic analyses result in a variant of unknown significance in a gene with either known or unknown classification. If the gene is inaccurately classified, it can lead to incorrect diagnosis and treatment. Thus, the accurate and standardized classification of genes and genetic variants is imperative for advancing both clinical practice and scientific research.”

To ensure the robustness of the methodology, a comparison with other databases, such as ClinGen and GenCC, was conducted to assess the results in light of currently used standards. High concordance was observed between the databases, indicating the success of the new approach.

The proof-of-concept study has created a genetic evidence database that can be continuously and rapidly updated as new gene–disease relationships are established. “This will result in the resolution of genes of unknown significance (GUS) into more defined categories, such as the limited, strong, and definitive designations,” predicts Kiel. “As more GUS are resolved, the diagnostic workup and patient diagnosis rates are improved.” 

The study could prove particularly beneficial to patients with rare or unassigned genetic diseases. Kiel concludes, “This study will help clinical labs and clinicians in determining which genes should be tested and which should be omitted from testing for a particular disease, avoiding unnecessary testing, inaccurate diagnosis, and treatment.”

As to future developments, further AI capabilities will be implemented to the system to enhance expert human curation efficiency and the user experience.