Fig. 2: Transcriptional phase inference with DeepCycle.
From: Cell cycle gene regulation dynamics revealed by RNA velocity and deep-learning
A Single-cell RNA-seq combined with RNA velocity analysis allows the detection of transcriptional changes within a single cell. B Fully circular RNA velocity patterns can be mapped to an angle describing the transcriptional state of a gene. By generalizing this to all genes, the angle will describe the actual transcriptional state of a cell. The angle is called the transcriptional phase. C DeepCycle infers the transcriptional phase of each cell. It takes as input the spliced-unspliced reads (si, ui) for a set of n genes (i = 1,…,n). By fitting the transcriptional phase θ, it can denoise and predict the unspliced-spliced expressions for each transcriptional phase. D Examples of cycling genes and fits of the RNA velocity patterns (red lines) in mESCs and human fibroblasts. E Cyclin E and B levels along the transcriptional phase in the three datasets. Negative controls for non-cycling genes have been added for comparison in Supplementary Fig. S10.
