“As on 4th June 2024 in La Jolla Institute for Immunology, scientists have developed a computational method linking DNA molecular marks to gene activity, a significant step towards machine learning to understand gene expression and disease development.”
Researchers at LJI have developed a graph neural network to process genomic data and identify enhancers which fine-tune gene expression levels. The network modeled after brain neurons, can integrate 3D information such as DNA physical interactions within cells. The researchers hope to use this tool to prioritize DNA interactions within the genome.
Researchers have developed neural networks to understand the relationship between the presence of a DNA modification called 5hmC and gene expression activity. 5hmC, the “sixth letter” of the DNA alphabet, is associated with enhancer activity. The conversion of 5mC to 5hmC on cytosine is linked to enhancer activity. Previous studies found that 5hmC’s location in the genome varies depending on cell type and genes expressed. This distribution controls gene expression in different cell types, such as liver cells, lung cells, and brain cells. Understanding 5hmC’s distribution can help in identifying the cell type producing the DNA being studied.
Researchers also added a new method using graph neural networks to predict interactions between genes and enhancers using 5hmC. The method reveals connections between genes and 5hmC in distant regions of the genome, prioritizing regions with gene expression enhancement. The researchers hope to further explore 5hmC distribution in human cells, immune cells, and cancer cells from patients.
The new LJI method may help understand the genetic mechanisms driving cancer development and potentially lead to faster and more accurate cancer diagnoses. By analyzing 5hmC distribution, doctors could help more patients and detect cancers earlier.
