Skip to main content
. 2025 Aug 26;26(4):bbaf411. doi: 10.1093/bib/bbaf411

Figure 4.

Benchmarks on the paired CITE-seq dataset.

(a) The comparison of Recall@k metric between different models for k = 10, 20, 30, 40, and 50 for the CITE-seq dataset (gene expression and cell surface proteins). (b) Normalized Median Rank metric for different models. (c) Cell type accuracies between models that show the percentage of closest embeddings from different modalities having the same cell types. (d) Comparison of ASW score across models shows the clusters’ quality.