Modeling the impact of data sharing on variant classification

ObjectiveMany genetic variants are classified, but many more are designated as variants of uncertain significance (VUS). Patient data may provide sufficient evidence to classify VUS. Understanding how long it would take to accumulate sufficient patient data to classify VUS can inform many important decisions such as data sharing, disease management, and functional assay development. Materials and MethodsOur software models accumulation of clinical data and their impact on variant interpretation to illustrate the time and probability for variants to be classified when clinical laboratories share evidence, when they silo evidence, and when they share only variant interpretations. ResultsOur models show that the probability of classifying a rare pathogenic variant with an allele frequency of 1/100,000 (1e-05) from less than 25% with no data sharing to nearly 80% after one year when labs share data, with nearly 100% classification after 5 years. Conversely, our models found that extremely rare (1/1,000,000 or 1e-06) variants have a low probability of classification using only clinical data. DiscussionThese results quantify the utility of data sharing and demonstrate the importance of alternative lines of evidence for the interpretation of rare variants. Understanding variant classification circumstances and timelines provides valuable insight for data owners, patients, and service providers. While our modeling parameters are based on assumptions of the rate of accumulation of clinical observations, users may experiment with the impact of these rates by downloading the software and rerunning the simulations with updated parameters. ConclusionThe modeling software is available at https://github.com/BRCAChallenge/classification-timelines.