Home Department: undeclared
Mentor: Anshul Kundaje (Genetics and Computer Science)
“Developing the Architecture of BPNet: A Deep Neural Network Designed to Discover Genomic Regulatory Elements”
Transcription factors (TFs) are proteins that bind to genomes to regulate the expression of genes. Understanding the landscape of TF-binding interfaces is crucial to understanding disease. For example, the binding interface of TFs androgen (AR) and glucocorticoid (GR) can elucidate GR allowing functional substitution for AR in prostate cancers. A popular method of exploring binding interfaces is SelexGLM, which fits a feature-based linear model to binding interface sequences. Anudeep is developing a cost-efficient neural network that can replace the mathematical analysis of SelexGLM. This enhancement could allow for increased and more advanced investigation of binding interfaces central to diseases.