Hunting for disease-causing genetic variants in non-coding DNA

2023

Challenge

Many diseases are caused by the improper regulation of genes. In complex diseases like type 2 diabetes, a large number of genes are involved. Although studies have made progress in identifying genomic variations linked to disease, it is still challenging to pinpoint the exact genetic changes responsible. This difficulty is heightened when these changes are in non-coding parts of the DNA, which don't directly produce proteins but instead control gene activity.

Advance

A team of IRP researchers, in collaborations with scientists at the University of Connecticut and Weill Cornell Medicine, utilized an artificial intelligence technique called deep learning to create a tool called TREDNet, which uses data from large studies to identify regions of DNA that actively participate in gene regulation. The tool then predicts how changes in these segments of DNA affect gene activity. Applying TREDNet to a large genetic study related to type 2 diabetes resulted in identifying four times as many potential disease-causing genetic variants than was known before. 

Impact

TredNET can be applied to identify genetic variants that contribute to a wide variety of complex human diseases and aid in the development of new treatments by uncovering biological processes that are disrupted by those genetic variants. Already, IRP researchers have also used the tool to uncover genetic variants strongly associated with autism spectrum disorders. Future planned applications include using it to uncover genetic variants related to the risk for Alzheimer's disease, Parkinson's disease, polycystic ovary syndrome (PCOS), glaucoma, and several other diseases.

View All Health Topics