Fig. 5: DOLPHIN captures biologically relevant alternative splicing events that signify cell-type-specific differences.

a Dot plots showing the PSI values of the top differentially spliced events identified by DOLPHIN. b GO biological process (GOBP) enrichment analysis of biologically significant differentially spliced genes identified by DOLPHIN, with alternative splicing-related terms highlighted in red. Adjusted P-values for each enrichment term were calculated using one-sided hypergeometric tests, followed by multiple testing correction using the Benjamini–Hochberg method. c Schematic illustration explaining PSI distribution splicing modality categorization. d PSI distribution for a single alternative splicing event, categorized by splicing modality across cell types, demonstrating that DOLPHIN provides clearer distinctions of splicing differences that align with cell type identities. e Splicing modality composition across single cells shows that DOLPHIN captures more distinct and biologically relevant splicing patterns by reducing the proportion of multimodal (null) categories, which represent PSI distributions without clear features. This demonstrates that DOLPHIN reduces ambiguity in alternative splicing event detection, enabling more precise analysis. f UMAP plots of cell clusters using PSI modality one-hot encoding demonstrate that the PSI splicing modalities identified by DOLPHIN retain strong cell-type-specific signals. DOLPHIN enhances the resolution of these cell-type-specific splicing patterns, providing clearer separation and biologically meaningful clustering compared to single-cell data alone. These biologically relevant alternative splicing events can contribute to more accurate cell type classification and offer insights into cellular diversity and potential disease mechanisms. P values from one-sided Student’s t-tests: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s. not significant. Exact P values are provided in the source data. Source data are provided as a Source Data file.