The new computer program "learns" to identify mosaic mutations that cause disease

Genetic mutations cause hundreds of incurable diseases
.
Among them, DNA mutations in a small percentage of cells are called mosaic mutations because they are present in a small percentage of cells, so they are very difficult to detect
.

Current DNA mutation software detectors do not do a good job of identifying mosaic mutations
hidden in normal DNA sequences when scanning the 3 billion bases of the human genome.
Often, medical geneticists have to examine DNA sequences with their eyes to try to identify or confirm mosaic mutations — a time-consuming job full of the potential for
error.

In this article, published on January 2, 2023 in Nature Biotechnology, researchers at the UC San Diego School of Medicine and the Reddy Children's Genomic Medicine Institute describe a way
to teach computers how to use artificial intelligence methods known as "deep learning" to find mosaic mutations.

Deep learning, sometimes referred to as artificial neural networks, is a machine learning technique that teaches computers to do what humans are born with: lead by example, especially learning
from large amounts of information.
In contrast to traditional statistical models, deep learning models use artificial neural networks to process visually represented data
.
These models function similarly to human visual processing, with greater accuracy and attention to detail, leading to significant advances in computational power, including mutation detection
.

"An example of an unresolved disease is focal epilepsy," said senior study author Joseph Gleason, MD, the Rady Professor of Neuroscience at the UC San Diego School of Medicine and director of
neuroscience research at the Rady Institute for Children's Genomic Medicine.

"Epilepsy affects 4 percent of the population, and about a quarter of focal seizures do not respond to
common medications.
These patients usually require surgery to remove the short-circuited part of the brain to stop the seizures
.
In these patients, mosaic mutations within the brain can cause epileptic foci
.

"We have a lot of people with epilepsy and we can't find out the cause, but once we apply what we call 'deep mosaic' to the genomic data, the mutations
become apparent.
" This has allowed us to improve the sensitivity of DNA sequencing for certain forms of epilepsy and has led to discoveries that point to new ways
to treat brain diseases.
”

Gleason said precise detection of mosaic mutations is the first step
in medical research to develop treatments for many diseases.

Dr.
Yang Xiaoxu, a postdoctoral scholar in Gleeson's lab, said DeepMosaic was trained on nearly 200,000 simulated and biological variants throughout its genome until "eventually, we were satisfied with
its ability to detect variants from data it had never encountered before.
" ”

To train the computer, the authors provide reliable examples of mosaic mutations as well as many normal DNA sequences, and teach the computer to tell the difference
.
By repeatedly training and retraining increasingly complex datasets and choosing between more than a dozen models, the computer was finally able to identify mosaic mutations
better than the human eye and previous methods.
DeepMosaic has also been tested on several never-before-seen independent large-scale sequencing datasets, outperforming previous methods
.

"DeepMosaic surpasses traditional tools in detecting mosaicism from genomes and exon sequences," said co-first author Xin Xu, a former undergraduate research assistant at the University of California, San Diego School of Medicine and now a research data scientist
at Novartis.
"The salient visual features captured by deep learning models are very similar
to what experts focus on when manually examining variables.
"

DeepMosaic is free and open
to scientists.
This is not a single computer program, but an open-source platform that allows other researchers to train their own neural networks to achieve more targeted mutation detection
using similar image-based setups, the researchers said.