FREMONT, CA – Researchers from Princeton University Lewis-Sigler Institute for Interactive Genomics recently utilized an artificial intelligence (AI) system to detect genetic anomalies in autism spectrum disorder. They initially analyzed the genomes of 1,790 families with a simplex autism spectrum disorder. An AI system was used to sort through 120,000 genomic disorders detected in the participants and separate the ones affecting the behavioral genes of the people with autism.
The AI system leveraged its deep learning capability to conduct multiple layers of analysis to detect the patterns which would otherwise be indiscernible to human analysts. The system taught itself to identify the biologically relevant sections of DNA and decide their role in the protein interactions that impacted gene regulation. The system also determined the effects of disrupting a single pair of DNA units on the protein interactions.
The algorithm analyzed all the chemical pairs in context to the other 1000 pairs surrounding it until it identified every mutation. The process showcased the capability of the AI system to predict the effects of altering the chemical pairs in the genome. The system revealed a list of DNA sequences likely to control genes that may interfere with the gene regulation.
Not more than 30 percent of the autism patients were aware of the genetic cause for their diagnosis before the study. The predictive ability of the AI proved crucial in identifying the mutations. Previous studies into the genetic causes of autism had failed to recognize the difference in the number of variations in the behavior-regulating genes of autism contracting patients when compared to neurotypical people.
The AI system employed by the Princeton researchers assessed the genetic mutations likely to cause a functional impact on behavior and identified a significant difference in number in the affected people. The data gathered from the study could prove vital to the family, friends, and physicians of the autism affected people as it will prevent them from making general classifications about their conditions.
No two cases of autism are alike, in terms of behavior or genetics. The results of the study show that the mutation in regulatory genes causes complex conditions in autism patients. The researchers are confident that their algorithm can be generalized to identify the role of genetics in different medical conditions. It could also be leveraged in treating neurological disorders, cancer, heart diseases, and other similar conditions.