A group of Belgian scientists have applied deep learning methods to investigate the mechanisms of gene regulation in human brain cells and chicken cells. Published in the journal Science, their work reveals that certain types of cells have remained unchanged over millions of years of evolution, while others exhibit significant differences between species.
Despite sharing the same DNA, brain cells are able to perform different functions due to a complex system of regulatory switches that control genes. These switches create a unique regulatory code, dictating which genes are activated in each cell. Professor Stein Aertes and his team from the VIB-KU Leuven center utilize machine learning methods to analyze these codes and their impact on brain evolution and disease development.
The researchers have developed algorithms that can identify both conservative and variable elements of the regulatory code in mammalian and avian brains. They discovered that in certain bird species, the regulatory codes closely resemble those found in the deep layers of mammalian neocortex, suggesting a potential evolutionary connection.
These advancements in research are not only valuable for evolutionary studies but also hold promise for medical applications. Previously, the Aertes group utilized similar algorithms to study melanoma mechanisms in mammals and fish. Now, they are aiming to leverage their models to enhance understanding of the genetic basis of neurological disorders like Parkinson’s disease.
Expanding their analysis to include a variety of animal species such as fish, deer, hedgehogs, and capybaras, the scientists aim to further explore brain evolution and develop novel approaches for diagnosing and treating genetic disorders.