Clinical genomic sequencing evolving at life-saving speed
The latest in genomic sequencing technology gives scientists access to the full spectrum of genetic variation for patient assessment. In practical application, the process allows healthcare providers to quickly find serious genetic conditions and initiate intervention.
In a recently published study, the Stanford University School of Medicine reported on use of sequencing technology combined with data analytics to identify suspected disease-causing variants in newborn intensive care unit cases. In the most quickly completed cases, the method identified a likely disease-causing variant about 40% faster than prior results.
In one case, a three-month-old, full-term infant presented in seizure. Magnetic resonance imaging (MRI) of the brain revealed no abnormalities. However, genetic diagnosis—completed in less than 8.5 hours—identified a variant and gene known to cause a neurodevelopmental disorder with early-onset epilepsy.
“This result halted further planned diagnostic testing, facilitated disease-specific counseling and prognostication, and aided in management of epilepsy by providing insight about reported seizure types and treatment response to common antiseizure medications,” according to Stanford researchers. In contrast, an epilepsy gene panel ordered at the patient’s presentation returned results two weeks later and showed only “multiple nondiagnostic variants of uncertain significance.”
Technology behind the scenes
Google Health and Pacific Biosciences (PacBio), a developer of genome-sequencing platforms, have partnered on work to advance genomic technologies as demonstrated at Stanford. PacBio says its technology can capture data from native DNA or RNA molecules to produce long-read sequencing at greater than 99.9% single-molecule accuracy.
Google worked with the University of California, Santa Cruz, Genomics Institute to build a method called DeepVariant that can analyze data for the fastest commercially available sequencing technologies. Google says that by combining DeepVariant with PacBio’s platform, researchers have been able to accurately identify diseases that would be difficult to diagnose with alternative methods.
Related story: HL7 FHIR to launch genomic data interoperability community
Further, under terms of the collaboration, Google’s genomic analysis, machine learning, and algorithm development tools will be used to improve PacBio’s variant calls, yielding additional insights from sequencing data. A 2021 study showed that Google’s DeepConsensus machine learning tool could increase the yield of 99.9% accurate reads by as much as 27% per instrument run.
“These improvements in accuracy and data analysis have the potential to enable more customers to experience the benefits of long-read sequencing as part of their research and translational projects, ultimately making a positive impact on implementing genomics in precision health,” said Christian Henry, president and CEO of PacBio.
Related story: Can synthetic patient data improve training for AI algorithms?
Frank Irving is a Philadelphia-based content writer and communications consultant specializing in healthcare and technology.