Fig. 3
From: A deep generative model for multi-view profiling of single-cell RNA-seq and ATAC-seq data
scMVP identifies accurate cell clusters from joint profiling cell line data. a–c Visualization of algorithms latent embeddings of three groups, algorithms specifically for joint modality datasets (shown as “Paired only”), algorithms of single omic (shown as “Single view”), and algorithms designed from both paired and unpaired datasets (shown as “Universal”) a UMAP visualization of scMVP, scVI, cisTopic, WNN, MultiVI, Cobolt, MOFA+, and scAI by Seurat v4 on the sci-CAR cell line dataset of A549, 293T, and 3T3 cells. b UMAP visualization of scMVP, scVI, cisTopic, WNN, MultiVI, Cobolt, MOFA+, and scAI by Seurat v4 on the Paired-seq cell line dataset of HEK293 and HepG2 cells. c UMAP visualization of scMVP, scVI, cisTopic, WNN, MultiVI, Cobolt, MOFA+, and scAI by Seurat v4 on the SNARE-seq cell line dataset of H1, BJ, K562, and GM12878 cells. d ARI scores for clustering on latent embeddings of benchmark algorithms