Transcriptional networks operate dynamically in vivo, but capturing and modeling these dynamics is an experimental and computational challenge. This presentation focuses on time building predictive network models based on time-series transcriptome data, and perturbing transcription networks in time. The outcome is a dynamic hit-and-run transcription model with relevance across eukaryotes.
In this talk, Dr. Gloria Coruzzi will probe dynamic transcription networks, computationally and experimentally. Using a machine learning approach called Dynamic Factor Graph, fine-scale time-series transcriptome data is used to infer network models that were validated both in silico using left-out data, and experimentally. To explore the molecular basis for underlying dynamic transcription, a cell-based assay was developed to follow the mode-of-action of a transcription factor (TF) within 1 minute of nuclear entry. This uncovered genome-wide support for a hit-and-run mechanism of transcription, in which de novo transcription initiated by a transient TF hit persists after the TF has run.