Autism in the "trash"
New causes of autism found in "junk" DNA
Using the artificial intelligence training method, scientists from the United States studied the genomes of 1,790 people with autism and found that mutations in the so-called "junk" DNA can cause this disorder, the press service of the Simons Foundation (New Causes of Autism Found in 'Junk' DNA) reports. The results of the study are published in the journal Nature Genetics (Zhou et al., Whole-genome deep-learning analysis identifies contribution of noncoding mutations to autism risk).
An analysis by a team from Princeton University and Rockefeller University predicted the consequences of genetic mutations in parts of the genome that do not encode proteins – these areas are often characterized as "junk" DNA. The number of autism cases associated with non-coding mutations was comparable to the number of cases associated with protein-coding mutations that disable gene function.
Only 1-2% of the human genome consists of genes that encode "blueprints" to create proteins. These proteins perform tasks throughout the body, such as regulating blood sugar levels, fighting infections, and sending messages between cells. However, the remaining 98% of our genome is also not a "dead weight". Non-coding regions help regulate when and where genes form proteins.
Mutations in protein-coding regions account for no more than 30% of autism cases in people without a family history of autism. Evidence suggests that autism-causing mutations should occur elsewhere in the genome. And they are not only inherited, but also arise spontaneously.
It is quite difficult to reveal which non-coding mutations can cause autism. One person can have dozens of them, most of which will be unique. Therefore, it is not possible to identify a common mutation by the traditional method.
In a new study, scientists have chosen a new approach. They trained a neural network to predict how a certain sequence will affect gene expression. The "smart" network tells you what any particular mutation is going to do, even one that is rare or has never been observed before.
Researchers have studied the genetic basis of autism by applying a machine learning model to a treasure trove of genetic data called the Simons Simplex Collection (Simons Foundation). The collection of the foundation contains the entire genomes of almost 2000 "quartets" consisting of a child with autism, a healthy sibling and their healthy parents.
These four had no previous family history of autism, which means that non-hereditary mutations were probably responsible for the condition of the sick child. (Such mutations occur spontaneously in sperm and eggs, as well as in embryos.)
The researchers used their model to predict the effect of non-inherited, non-coding mutations on each child with autism. Then they compared these predictions with the effects of the same, unchanged, chain in a healthy sibling of a child.
The analysis suggests that non-coding mutations in many children with autism alter gene regulation. Moreover, the results showed that mutations affect the expression of genes in the brain and genes already associated with autism – for example, those responsible for the migration and development of neurons. This explains why autism is most likely to manifest in the brain, the researchers note.
The significance of this work goes beyond the study of autism. Scientists can use the same methods used in the new study to study the role of non-coding mutations in diseases such as cancer and heart disease, says co-author of the study Gian Zhou from Princeton: "This allows us to take a fresh look at the cause of not only autism, but also many human diseases."
Portal "Eternal youth" http://vechnayamolodost.ru