A study has used artificial intelligence (AI) to detect genetic abnormalities in relation to autism spectrum disorder.
Princeton University’s Lewis-Sigler Institute for Interactive Genomics conducted the research, analysing the genomes of 1,790 families with simplex autism spectrum disorder, where only one child has the disorder and no other family members do.
The AI system performed continual layers of analysis, picking out patterns that humans could not. The system then taught itself how to identify biologically relevant segments of DNA, predicting whether they affected gene regulation.
Less than 30 per cent of participants with autism had a genetic cause that had been previously identified before the study, but it is hoped that the new mutations that have been found will increase that figure.
Researchers praised the predictive ability of the AI, with previous studies of the genetic cause of autism failing to identify any notable difference in the number of mutations in the regulatory genes when compared with people who do not have autism.
Chandra Teesfeld, Researcher at Princeton University, said: “They say that when you meet one person with autism you have met one person with autism because no cases are alike. Genetically, it seems to be the same way.”
The algorithm works by ‘sliding along the genome’, analysing every chemical pair in relation to those around it, until it has scanned all of the mutations.
Because of this, the system can predict the effect that mutating every chemical unit would have. Then it reveals an ordered list of DNA sequences that are expected to regulate genes and mutations.