Generalizing RNA velocity to transient cell states through dynamical modeling
Bergen V, Lange M, Peidli S, Wolf FA, Theis FJ. Generalizing RNA velocity to transient cell states through dynamical modeling. Nat Biotechnol. 2020 Dec;38(12):1408-1414. doi: 10.1038/s41587-020-0591-3. Epub 2020 Aug 3. PMID: 32747759.
RNA velocity is the change in mRNA abundance. This depends on the ratio between spliced and unspliced mRNA and the assumption that they reach a steady-state. Velocities are determined as the deviation of the observed ratio from the steady-state ratio. This depends on the gene-level capture of the full splicing dynamics with transcriptional induction, repression and steady-state mRNA levels and on all genes sharing a common splicing rate. While RNA velocity prediction allows us to see how cells change their state, the base algorithm does not do well when the assumptions of a common splicing rate or having the full splicing dynamics are violated. Further, not all experiments are done in a time-frame that involves reaching the steady-state.
scVelo uses a likelihood-based dynamical model rather than a steady-state model that solves the problems with those assumptions. It works by generalizing RNA velocity estimation to transient systems and systems whose subpopulation kinetics are heterogeneous. They can infer gene-specific reaction rates of transcription, splicing and degradation. They can also infer the underlying gene-shared latent time which represents the cell’s internal clock that describes the cell’s position in the underlying biological processes. This is different from existing pseudotime methods because it is based only on transcriptional dynamics and takes both speed and direction into account. Dynamical modelling allows for delineating cell cycling, lineage commitment, cell cycle exit and endocrine differentiation. This method, however, still assumes constant reaction rate and is limited by alternative splicing.