Penetrance and variant consequences—Two sides of the same coin?

Over the past few decades, sequencing a human genome has gone from a moonshot project to standard clinical practice, revealing thousands of variants in each individual. A crucial question in linking genome sequences to health outcomes is, which gene variants affect the risk of getting a particular disease, and if so, how many of the carriers go on to develop the disease? For example, a patient that carries a particular variant may benefit from more frequent screening for diabetes and breast cancer. In many such cases, early detection of disease onset can greatly improve overall outcomes for the patient. On page 894 of this issue, Forrest et al. (1) describe the integration of large-scale phenotype data, genetic sequence data, and artificial intelligence for approximating disease risk using machine-learning methods. This framework provides precedent for the synthesis of clinical and genetic data for personalized treatment over a variety of diseases. [ABSTRACT FROM AUTHOR]