AI algorithm analyses gene pair variants to improve rare disease diagnosis

Chloe Kent 6 June 2019 (Last Updated December 23rd, 2019 10:22)

A team of Belgian researchers are using artificial intelligence (AI) to identify the causes of rare disease and improve their diagnosis.

AI algorithm analyses gene pair variants to improve rare disease diagnosis
VarCoPP makes it possible to simultaneously test the combinations of different variants in gene pairs to predict their potential pathogenicity. Credit: Shutterstock

A team of Belgian researchers are using artificial intelligence (AI) to identify the causes of rare disease and improve their diagnosis.

The leam, led by Professor Tom Lenaerts of the Université Libre de Bruxelles, developed an AI algorithm called the Variant Combinations Pathogenicity Predictor (VarCoPP). VarCoPP makes it possible to identify combinations of genetic variants or abnormalities that cause rare diseases through computer analysis.

Around 80% of rare diseases are thought to be genetically determined, so it’s important for doctors to be able to predict which genetic variants in a patient’s genome may be responsible for their illness.

VarCoPP makes it possible to simultaneously test the combinations of different variants in gene pairs to predict their potential pathogenicity.

The team trained VarCoPP using the DIDA database of rare diseases  that they developed in 2015, as well as data from the 1000 Genomes Project.

The tool has been validated on 23 independent pathogenic gene combinations, delivering confidence intervals of 95% and 99% to help doctors zoom in on the most important predictions.

This demonstrates that VarCoPP picks up pathogenic combinations accurately and precisely.

The researchers are now attempting to use the results to identify the genetic causes of rare diseases in patients for whom no cause has previously been identified.

The team has introduced an online diagnostic platform for clinicians and researchers based on VarCoPP, known as the Oligogenic Resource for Variant AnaLysis (ORVAL). ORVAL is dedicated to identifying networks of patient pathogenic variant combinations with the aim of uncovering the root causes of a disease for patients who cannot rely on traditional diagnostic pathways.

VarCoPP and ORVAL provide the opportunity to study genetic variant combinations associated with disease where different combinations of genetic variations are likely to be the cause, whether the causal genes are known or unknown.

This could range from examining the 20 genes associated with the rare Bardet-Biedl syndrome or the hundreds of genes associated with autism.

A deep learning AI has recently been used by researchers at Princeton University to identify a number of previously overlooked mutations in the DNA of autistic people.