Fig. 6

Comparisons with multiomics data integration methods. a t-SNE visualizations of scRNA and scATAC-seq data from co-assayed A549 cells, colored by measurements (RNA vs. ATAC) after integration with Seurat (left) and LIGER (middle). Right panel: comparisons of alignment score (quantified by the entropy of batch mixing) from perfect alignment (termed as gold-standard) with that computed from the aligned t-SNE space using Seurat and LIGER. p values are from the Wilcoxon rank-sum test. b Cells are colored by the data collection times. Right panel: comparisons of silhouette coefficient computed from the t-SNE coordinates of each cell generated by scAI with that computed from the aligned t-SNE space using Seurat and LIGER. c, d UMAP visualizations of scRNA and scATAC-seq data from co-assayed mouse kidney cells colored by measurements (RNA vs. ATAC) (c) and published cell labels (d) after integration with Seurat and LIGER. The alignment score and silhouette coefficient were also shown