The expression of genes is controlled by gene regulatory networks, which are complex networks that capture the interactions between genes, non-coding regulatory elements, and transcription factor proteins. Understanding gene regulatory networks has wide applications in disease treatment since many disease-related genetic mutations are located in the non-coding regions of our genome and indirectly alter gene expression. Though we understand how some pairs of regulatory elements and genes interact, many interactions still remain unknown. The goal of this project is to discover these interactions by architecting a Graph Neural Network (GNN) based deep-learning method, which takes single-cell gene expression and chromatin accessibility data as inputs.

Having taken algorithms and machine learning courses and being interested in the applications of computer science in human health, I believe this SuperUROP is an excellent opportunity to investigate a fundamental problem in genetic mechanisms using computational methods. While learning about regulatory networks, I am excited to gain more experience in breaking problems into subproblems, handling ambiguity, and communicating my work to others.