LIET Model: Capturing the kinetics of RNA polymerase from loading to termination

Transcription by RNA polymerases is an exquisitely regulated process and the principle step of the central dogma. Transcription is the primary determinate of cell-state and most cellular perturbations (e.g. disease, drug treatment, or abiotic stressors) impact transcription by altering the activity of polymerases. Thus, the ability to detect and quantify detailed changes in polymerase activity would yield considerable insight into most cellular processes. Nascent run-on sequencing assays provide a direct readout of polymerase activity, but no tools exist that comprehensively model this activity at genes. Thus, we focus on RNA polymerase II (RNAP2), which is responsible for transcribing all protein-coding genes and many ncRNAs. Here we present the first model to fully capture the complete process of gene transcription by RNAP2. For an individual gene, this model parameterizes each distinct stage of RNAP2 transcription— Loading, Initiation, Elongation , and Termination , hence LIET—in a biologically interpretable way, and combines them in a Bayesian mixture model intended to be applied to nascent run-on sequencing data. Additionally, we provide a user-friendly, open-source software implementation of the model, which we use to demonstrate its performance. With our improved modeling of Loading and Initiation we demonstrate that these processes are characteristically different between sense (gene coding) and antisense (PROMPT) strands. Furthermore, our model is the first to robustly quantify the Termination phase of transcription. By applying the LIET model to 24 different human cell-types, our analysis indicates that the position of dissociation appears to be highly consistent across cell-types, indicative of a highly regulated process. Furthermore, by applying the LIET model to perturbation experiments, we demonstrate its ability to detect specific regulatory changes in pausing ratio (at 5 ^′ end), strand-bias, and dissociation location (at 3 ^′ end). This opens the door to differential assessment of transcription regulation at the level of not only individual genes but also individual transcription stages within a gene.