Dec 13, 2018

Mapping the brain to see how diseases may start

Illustration of brain with letters of DNA coding arising off of it

Illustration: Aïda Amer/Axios

By examining brain tissues and developing an AI that scours a huge and improved gene database, research is narrowing in on the genes underlying serious psychiatric disorders, a consortium of scientists announced Thursday.

Why it matters: Ultimately, they want to determine how psychiatric disorders form in the brain and to develop more targeted drugs or other treatments. The results from the consortium of 15 research institutions (called PsychENCODE) and organized by the National Institute of Mental Health (NIMH), are considered a "big step" toward reaching that ultimate goal, particularly for schizophrenia, bipolar disorder and autism, some experts tell Axios.

Background: It's now known there are hundreds of genes in the brain that can play a role in boosting the risk factors for these psychiatric disorders — but how the disorders develop on a molecular level, and what role those variations in genes play, is not known.

  • "What we didn't know, is what actually happens in the brain," Thomas Lehner, director of the NIMH Office of Genomics Research Coordination, describes in a YouTube video. "Some areas are more likely to be involved in the expression of a disorder than others."
  • This project targeted the non-coding part of the genome, since prior research linked that region to certain disorders but not much was known about it.

What they did: The 10 studies, published in Science, Science Advances and Science Translational Medicine, looked at about 2000 individual brain samples — both tissues and single cells.

  • Several documented the various stages of brain development across the human life span, including the rarely examined stage of a fetal brain.
  • Researchers developed deep-learning algorithms to process the huge amount of data and predict genetic risk for disorders.

The deep-learning model that examined and integrated the huge amount of data to predict risk factors for disorders was lauded by several sources as an important advance.

  • It was 75% successful in estimating the risk factors, which is significantly higher than current methods, according to Yale's Mark Gerstein, who is an author of several of the studies.
  • "It actually told us which pathways and genes and whatnot were implicated in this particular disorder," Gerstein tells Axios, adding that it's obviously not ready for clinical use yet.
  • "It's nowhere near what doctors need, which is 100% ... but we are inching toward using these data sets to make better diagnoses in the first place," Mount Sinai's Schahram Akbarian, another author of several studies, tells Axios.

Other findings pointed out by some experts include:

  • Lower levels of inhibitory neurons were found in autistic brains while having greater amounts of excitatory neurons, Gerstein says.
  • The expression of a gene associated with schizophrenia was found to involve another gene (POU3F2) that's been linked to bipolar disorder.

The big picture: A lot of new research is expected to flow from these findings."The biggest impact will be [the data] as a scientific resource for the whole community," Stanford University's Michael Snyder, who was not part of these studies, tells Axios.

Go deeper:

Go deeper