Scientists are using machine learning to decode gene regulation in the developing human brain

In a scientific feat that advances our understanding of the genetic changes that shape brain development or lead to mental disorders, a team of researchers combined high-throughput experiments and machine learning to analyze more than 100,000 sequences in human brain cells and identify more than 150 variants that likely cause disease .

The study, conducted by scientists from the Gladstone Institutes and the University of California, San Francisco (UCSF), establishes a comprehensive catalog of genetic sequences involved in brain development and opens the door to new diagnostics and treatments for neurological conditions such as schizophrenia and autism spectrum disorders. The journal publishes an article entitled “Massively Parallel Characterization of Regulatory Elements in the Developing Human Cortex” Science

“We collected a huge amount of sequence data in non-coding regions of DNA that were already suspected to play a large role in brain development or disease,” says senior researcher Dr Katie Pollard, who is also director of the Gladstone Institute for Data Science and Biotechnology.

“We were able to functionally test over 100,000 of them to find out whether they affected gene activity and then pinpoint sequence changes that could alter their activity in disease.”

Pollard co-led this large-scale study with Nadav Ahituv, Ph.D., professor in the Department of Bioengineering and Therapeutic Sciences at UCLA and director of the UCLA Human Genetics Institute. Much of the experimental work on brain tissue was conducted by Dr. Tomasz Nowakowski, associate professor of neurological surgery at the UCSF Faculty of Medicine.

In total, the team found 164 variants associated with mental disorders and 46,802 sequences with enhancer effects in developing neurons, meaning they control the function of a given gene.

These “enhancers” could be used to treat mental illnesses in which one copy of the gene is not fully functional, Ahituv says. “Hundreds of diseases result from the malfunctioning of a single gene, and it may be possible to use these enhancers in treatments to make them do more.”

The focus is on organoids and machine learning

In addition to identifying disease-related enhancers and sequences, the study has implications in two other key areas.

First, the researchers repeated part of the experiment using a brain organoid developed from human stem cells and found that the organoid effectively replaced the original. Notably, most of the genetic variants detected in human brain tissue replicate in the brain organoid.

“Our organoid compares very well with the human brain,” says Ahituv. “As we expand our work to test more sequences for other neurodevelopmental diseases, we now know that the organoid is a good model for understanding gene regulatory activity.”

Second, by feeding vast amounts of DNA sequence and gene regulatory activity data into a machine learning model, the team was able to train the computer to successfully predict the activity of a given sequence. These types of programs can enable “in silico” experiments, which allow scientists to predict the results of experiments before they are performed in the laboratory. This strategy enables scientists to make discoveries faster and with fewer resources, especially when large amounts of biological data are involved.

Dr. Sean Whalen, senior research scientist at Pollard Lab in Gladstone and co-author of the study, says the team tested the machine learning model using sequences from model training to see if it could predict the gene expression activity results already collected.

“The model had never seen data like this before and was able to make predictions with a high degree of accuracy, showing that it had learned general principles about how non-coding regions of DNA influence genes in developing brain cells,” Whalen says. “You can imagine how this could open up many new possibilities in research and even predict how combinations of variants might function together.”

A new chapter in brain discoveries

The study was completed as part of the PsychENCODE consortium, which brings together multidisciplinary teams to generate large-scale regulatory and gene expression data from human brains across several major psychiatric disorders and stages of brain development.

By publishing multiple studies, the consortium aims to shed light on poorly understood mental health conditions, from autism to bipolar disorder, and ultimately pioneer new approaches to treatment.

“Our study contributes to this growing body of knowledge by demonstrating the utility of using human cells, organoids, functional screening methods, and deep learning to investigate regulatory elements and variants associated with human brain development,” says Chengyu Deng, Ph.D. postdoctoral researcher at UCSF and co-author of the first study.