Fig. 3From: Single-cell multi-omics integration for unpaired data by a siamese network with graph-based contrastive lossMinNet batch effect removal outperforms other algorithms. While separating cell types and mixing modalities, our model showed the best performance in removing batch effect, the most common challenge in integrating different omics data from distinct sources. A UMAP visualization of the embedding space generated by all algorithms. B Silhouette score indicates that while separating cell types, our model mixes batches well. C Label transfer accuracy from one donor’s transcriptome data to another donor’s chromatin accessibility dataBack to article page