topicvelo 0.0.2

TopicVelo: Dissection and Integration of Bursty Transcriptional Dynamics for Complex Systems

TopicVelo is a novel approach for RNA velocity inference in general systems, including immune 
response studies. It infers the cells and genes associated with distinct active processes 
via probabilistic topic modeling, and uses these to estimate process-specific velocity 
parameters and transition probabilities, which are then integrated into large-scale transition
matrices. Parameter accuracy is also improved by efficiently fitting unsmoothed counts to a 
transcriptional burst model. In biologically varied datasets, this approach outperformed the 
state-of-the-art method, recovering parameters and transitions that were better experimentally 
supported or recovered previously only with the aid of metabolic labeling or multiple time points.

For more information please see our preprint
(https://www.biorxiv.org/content/10.1101/2023.06.13.544828v1.full)