Diploid, a Leuven-based genomics startup, announces the release of Moon, the first software package that autonomously diagnoses rare diseases using artificial intelligence. Moon reduces the time for genome interpretation from hours or days to mere minutes, making it the world’s fastest software for rare disease diagnostics. Moon has been tested by the University of Antwerp (UZA) in Belgium and by the National Institutes of Health (NIH) in Bethesda, USA.
Faster than geneticists
Moon was used to analyse cases that had been previously diagnosed by highly skilled geneticists. In 90% of these patients, Moon comes up with the correct diagnosis and in 97% of the patients, the disease causing variant is in the top 3 of Moon’s list of candidate variants. In addition, Moon managed to reach a diagnosis for patients who remained undiagnosed after manual interpretation.
Overall, the diagnostic yield is on a par with the work of an expert geneticist, while the processing time is significantly shorter. The software only requires the input of the patient’s symptoms, age, gender, and sequencing data (WES or WGS). Using proprietary artificial intelligence algorithms, Moon generates a diagnosis within 3 minutes.
Peter Schols, CEO and founder of Diploid, says: “The fact that Moon can go from genetic data (VCF) to a diagnosis in mere minutes, instead of hours or days, enables next generation sequencing to be used in new domains. For example, for infants with genetic diseases in emergency care units, a fast diagnosis can make the difference between life and death.”
Infinite interpretation
Moon can operate fully unsupervised and is always up to date with the latest genetic discoveries. This allows for fast and affordable re-analysis of sequencing data from undiagnosed patients on a monthly or even weekly basis.
Prof.
Geert Mortier, chairman of the Department of Medical Genetics at Antwerp University Hospitals, one of the early beta testers of Moon, said: “The proof of the pudding is in the eating: we’ve sent solved cases to test the software and in each case the causal mutation was found in no time.”
Diploid is targeting bioinformatics companies who will be able to embed Moon into their existing variant analysis software, to add intelligence to their current diagnostic processes. Hardware companies looking to integrate NGS interpretation software into their products can also use Moon. Other potential customers include large-scale genome centres and country-wide genome programmes interested in integrating Moon’s analytical power into their existing pipelines.
