Dynamic PAs during cortical neurogenesis and benchmarking of Infernape. (A) UMAP showing main cell types in the E14.5 dorsal mouse forebrain (5482 single cells). Data were analyzed from a previous study (La Manno et al. 2021). (B) UMAP showing transcriptome-wide average WARM values across cells. Higher WARM values indicate higher usage of distal PAs. (C) Empirical cumulative density curves of average WARM values for neuron, RGC, and IPC. Cell type–specific WARM values are calculated for APA signals of all multiple-PA genes. Wilcoxon rank-sum tests were used to test the overall APA difference between cell types. (*) adj.P-value < 0.05, (**) adj.P-value < 0.01, (ns) nonsignificant. (D) Scatter plot of WARM values showing 3′-UTR lengthening in neurons. Each dot represents a transcript, and the x/y-axes represent WARM values for the two cell types in comparison. Significant and nonsignificant differential APA events are colored red and gray, respectively. (E) UMAP showing WARM values for the Gnb1:ENSMUST00000165335.7 transcript were higher in neurons than in RGCs. (F) Coverage plot of the Gnb1:ENSMUST00000165335.7 transcript for neurons and RGCs, from both scRNA-seq (blue and pink) and bulk RNA-seq data (gray). (G) Coverage plot of the Gnao1 gene for single-cell and bulk RNA-seq of neurons and RGCs. (H) RNA in situ hybridization (ISH) results of Gnao1 in the E15.5 mouse brain (coronal section). (I) RNA ISH results of Gnao1 (zoom-in for H): The distal PA (Neuron_probe) shows a higher signal in the cortical plate (CP) than in the ventricular zone (VZ), whereas the proximal PA (RGC_probe) shows a higher signal in the VZ than in the CP. (J) The number of peaks identified by different PA methods using the E14.5 scRNA-seq data. MAAPER was not included here because it outputs PA coordinates instead of peak coordinates. (K) The proportion of differential PA genes that is identified by at least one of the other single-cell PA methods. (L) The number of differential PA genes that are shared by at least one of the other single-cell PA methods.