Overview
- Researchers validated Talos on two previously analysed cohorts totalling 1,089 patients and recovered about 90% of known genetic diagnoses while returning an average of 1.3 candidate variants per family.
- In a larger study of 4,735 previously undiagnosed children and adults, Talos produced 241 new diagnoses, a 5.1% increase in diagnostic yield reported in Nature Medicine.
- The system translates newly published gene and variant findings into candidate diagnoses with a median turnaround of 32 days after the relevant knowledge appeared and in some cases within one day.
- Talos is open-source, auditable, designed to run on standard computing infrastructure at very low cost — the team estimates under US$12 to run the initial workflow for 1,000 genomes and under US$2 per genome per year for ongoing monthly reanalysis — and the tool has already prompted testing of more than 50 additional family members.
- The work highlights that genomic data can be stored and reinterpreted as science advances, offers a practical path for diagnostic labs to scale routine reanalysis, and sets the stage for wider adoption and future integration with artificial intelligence.