Myriad Genetics Uses DNAnexus on AWS to Bring Precision Medicine to the Clinic

Clinical genomic testing can help to guide patients and healthcare providers to make informed decisions for personalized care, but genetic tests require massive computational power and complex bioinformatics pipelines to ensure the speed and accuracy required for clinical use.

By applying artificial intelligence and machine learning models to the enormous amount of genome-sequencing data processed in its bioinformatics pipeline, Myriad Genetics “delivers the next generation in clinical genomics validity and utility,” says Kevin R. Haas, senior vice president of technology at Myriad Genetics. When Myriad runs a clinical test, its bioinformatics pipeline needs to “churn on that genome sequencing data to be able to produce a high-confidence call,” says Haas. This issues a report that says the tested individual’s results are either normal or that the person screened positive for certain genomic alterations that may impact their risk of developing cancer or indicate that they are carriers of genetic disease. Then, together with their care teams, patients can make decisions about their future healthcare.

Operating its bioinformatics pipeline with DNAnexus, an Amazon Web Services (AWS) Advanced Technology Partner in the AWS Partner Network (APN), Myriad Genetics processes petabytes of data and runs thousands of analysis iterations plus validation processes to check testing results. Each of those iterations can require massive computational power, using up to 25,000 cores. “You can see how we really need the scale to do this because it is a lot of cores, but we only need them for a brief period of time, so Myriad really benefits from the on-demand scalability of AWS,” says Haas. Using AWS Step Functions and AWS Lambda, Myriad Genetics can orchestrate its complex workflows and run code without provisioning or managing servers and pays only for the compute time consumed. The Myriad Genetics team also uses AI and machine learning powered by GPU instances on AWS to improve its testing speed and accuracy, enhance clinical interpretation, and discover novel indications and biomarkers to help it develop new products. Now, Myriad Genetics has “gone from weeks to do one single iteration to doing this in a matter of hours,” says Haas, meaning it can produce better, faster tests and bring new products to the table, which is important for the over one million patients that use its products annually.