Michael Beer, Ph.D.

Professor, Departments of Biomedical Engineering, Genetic Medicine and Oncology, Johns Hopkins University School of Medicine

Host: Roger Pique-Regi, PhD

“Sequence-based Machine Learning for Modeling Cell State Transitions in Development and Disease”

Abstract

I will describe DNA sequence-based machine learning models which identify the active TFs controlling human cell/tissue-specific enhancer activity. Using these models, we designed CRISPRi screens in the hESC to endoderm transition (Luo, … Huangfu, Beer Nat Gen 2023). We discovered multiple enhancers flanking each TF gene, but perturbation of individual enhancers altered the threshold for transition into endoderm, but not the final steady state expression levels. We developed dynamic gene regulatory network models which explain the weak effect of enhancer perturbation once cell state is established and suggest transition-based strategies to more successfully identify regulatory contributions to human disease.