This section describes several algorithms and methods that interpret and clarify genetic data and explains how these applications may arise in federal court cases. It also raises critical questions for federal judges to consider when they review these methods and tools.

In 2016, Amy Williams sued Quest Diagnostics/Athena—a medical testing service—alleging that Athena made an error when it performed a genetic test on her two-year-old son in 2007, and that this error contributed to the inappropriate treatment of his seizures, and ultimately to his death.[1]

The suit spotlighted a critical aspect of medical genetics: the evolving understanding of genetics will lead to later reevaluation of certain test results. Specifically, the Athena genetic test revealed Williams’s son carried a variant in a gene called SCN1A. Several different variants in the gene have been shown to result in seizures and epilepsy. But at the time of the test, Athena reported that the effect of the child’s specific variant on SCN1A was unknown and couldn’t be classified as contributing to his illness—a common outcome in medical genetic testing.

Years later, Athena reclassified the variant, now determining that it did affect SCN1A in a way that contributed to illness. Williams sued, claiming that the reclassification was based on research available at the time of her son’s test and that Athena had been negligent in ignoring it.

Williams v. Quest Diagnostics foregrounds several aspects of how genetic data are leveraged into useful and actionable tools. For the vast majority of the genome, we do not know the effect any specific variant or combination of variants will have on an individual’s health, appearance, or behavior. Yet, with the growing number of large genetic datasets, researchers' abilities to assign significance to variants whose clinical impact was previously unknown are consistently improving. Additionally, updated algorithms and tools, including those using artificial intelligence, can now predict a genetic variant’s clinical effect, as well as an individual's physical and behavioral traits.

Although predicting traits from genetic data has improved, these methods should still be used cautiously for many reasons. For one thing, many algorithms and methods are developed from datasets derived primarily from people of European ancestry, which dramatically impacts the tools’ predictive power when applied to people of other ancestries. Another reason is that interpretative tools—such as forensic probabilistic genotyping systems—may have been tested under very specific parameters, so their validation across a greater diversity of scenarios may be limited. Applying these tools and methods to situations for which their use has not been validated can result in unanticipated consequences and unpredictable biases.

As judges consider the applied use of these methods in cases before their courts, it is important to review the data used in any given method’s development, the protocols applied in its validation, and the ever-developing knowledge of genetics.