Lecture 7. Predicting Regulatory Genomic Sequences with Novel Deep Learning Approaches (ENG)

Ramzan UMAROV, PhD (RIKEN)

Lecture Abstract

Gene transcription regulation, undoubtedly one of the most important processes in molecular biology, requires transcription machinery to bind to the promoter of the gene and the stimulation from distal DNA regulatory regions called enhancers. Although enhancers can be located even one megabase away from the transcription start site of their target genes, underlying 3D chromatin structure allows enhancers to regulate transcription of the distal genes. Moreover, enhancers are highly enriched in disease-associated variants and thus deciphering the interactions between enhancers and genes is crucial to also understanding the molecular basis of genetic predispositions to diseases.

To address laborious and costly experimental validations of enhancer gene targets, computational methods have recently emerged as valuable alternatives in enhancer research. I will first briefly introduce principles of how enhancers regulate gene expression and experimental and computational approaches to discover enhancers and their target genes. Next, I will show how recent deep learning approaches take advantage of vast publicly available multiomics data to model gene expression, offering a new paradigm to study links between enhancers and their genes. I will also talk about our new AI method to study enhancers by analyzing gene expression models and how it can overcome some of the issues encountered in current state-of-the-art approaches.

In summary, this lecture will provide an overview of how we can better understand gene expression regulation using deep learning methods.