Search
  • News
  • Genetic Networks
  • Learning in Machines & Brains

New algorithm will help predict which genetic mutations lead to autism

by CIFAR May 29 / 14
autism_1280X420
Image above: These heat maps of the brain show the association between gene expression levels and the frequency of rare genetic mutations connected to autism in 16 brain regions at three stages of life. The redder the area, the greater the burden of damaging mutations. Image courtesy of Nature Genetics

Scientists have identified a critical set of genes that is likely involved in the development of autism, according to a paper published in Nature Genetics. They have also developed an algorithm that could help doctors predict whether mutations in one of these genes will actually lead to the disease.

CIFAR Senior Fellows Stephen Scherer (The Hospital for Sick Children), Brendan Frey (University of Toronto) and their team utilised large gene expression data sets from healthy individuals to identify 1,744 genes with a particular set of characteristics. The genes turn on at critical periods of brain development and express at high levels but without a high rate of mutation, suggesting they are critical to normal brain function.

When they compared the segments of these genes responsible for protein coding to those of the roughly 100 genes in which mutations are currently known to link to autism, they overlapped completely.

“We predict that the other 1,644 or so remaining genes are also cognition genes that will eventually be found to be involved in autism or related neurodevelopment disorders,” says Scherer, the lead author.

The new knowledge will help researchers connect the huge range of genetic mutations visible within individual autism patients to the larger autistic population as a whole. It could also help them better understand the strange clinical situations that sometimes arise, such as mutations within autism-related genes that don’t actually lead to the disease. Ideally, within these patterns will lay the key to predicting whether autism will develop.

“In our new study we’ve finally discovered a unifying set of characteristics in the DNA that we can weave into a ‘genetic formula’ that helps us calculate which genetic mutations have the highest probability of causing autism, and equally important, which alterations do not have a role,” Scherer said in a release by the Hospital for Sick Children.

Scherer says interdisciplinary collaboration was central to this breakthrough in understanding the genetic code of autism. Scherer, a genome scientist and a fellow in CIFAR’s Genetic Networks program, collaborated with computer scientist Frey, who is cross-appointed to Genetic Networks and Learning in Machines & Brains (formerly known as Neural Computation & Adaptive Perception).

“This paper embodies ideas of precisely the things we try to accomplish in our CIFAR network, that is, extract meaningful patterns of information from extremely large and complex datasets,” Scherer says.

The gene expression data sets were compiled from samples obtained from multiple brain regions and time points (pre-natal development, childhood and adulthood), in autistic and in normal brain functioning individuals. Scherer says his work on this project spanned more than a year, during which the presentations and discussions held at three CIFAR meetings helped his team solidify their methods and ideas.

“Without the CIFAR network I suspect we would have not been the first to come upon this idea and then take it to publication,” Scherer says.

“I do think this will be seen as one of the biggest advances from our laboratory in the past decade.”

This research was funded in part by the University of Toronto McGlaughlin Centre, NeuroDevNet, Genome Canada, the Ontario Genomics Institute, the Canadian Institutes for Health Research, the Canada Foundation for Innovation, the government of Ontario, the Ontario Brain Institute and Autism Speaks.