A newly developed open-source tool, Talos, has been shown to effectively detect new rare disease diagnoses by frequently and automatically reexamining stored DNA data, according to findings published on June 24. The research highlights that this approach could provide more timely answers for hundreds of families.
Talos was created and validated by researchers in Australia and the United States. According to the study, the tool identified new genetic diagnoses in over 200 patients where previous genomic testing had not found a cause for their condition. The results also suggest that Talos lays the groundwork for artificial intelligence-enabled approaches in genomic medicine.
The project involved collaboration between Murdoch Children's Research Institute (MCRI), Victorian Clinical Genetics Services (VCGS), Centre for Population Genomics, Broad Institute of MIT and Harvard, and Microsoft Research. Professor Zornitza Stark from MCRI said Talos would transform outcomes for patients and families affected by rare diseases. She said more than half of patients remain undiagnosed after initial genomic tests despite advances in technology.
Researchers first validated Talos using two previously analyzed cohorts involving 1,089 patients from the US and Australia. The tool successfully identified about 90 percent of known diagnoses while returning an average of just 1.3 candidate variants per family. Kaitlin Samocha from Broad Institute said, "As genetic sequencing becomes a standard part of healthcare, the backlog of undiagnosed families is growing rapidly. We designed Talos to return only a few variants per patient, reducing the analytical bottleneck and speeding up the time to diagnosis." In further testing with a cohort of 4,735 individuals with rare diseases who remained undiagnosed after earlier testing, Talos delivered an additional diagnostic yield of 5.1 percent across various conditions.
Professor Daniel MacArthur from Centre for Population Genomics said automated reanalysis allows knowledge about gene–disease associations to be translated into clinical benefit faster than traditional models, "Every year, hundreds of new gene–disease associations and thousands of new variant interpretations are published," he said.
Microsoft Research Principal Researcher Jeremiah Wander described Talos as open-source and auditable: "Talos is open‑source, auditable, and designed to run on standard computing infrastructure... The study provides critical evidence to inform future policy around rare disease diagnostics."
