Table 1 Cell type clustering and batch correction performance on integrating multiple scRNA-seq datasets. The clustering performance is measured by the normalized mutual information (NMI), where NMI_cluster=1 implies correct clustering by cell type annotations. (1−NMI_batch)=1 implies no batch effect present after integration. Hyperparameter m is the number of “ghost” cells, and d is the dimension of the reduced gene expression matrix. The bolded entries are the best performance metrics for each scenario. See the “Methods” section for details on NMI_cluster and NMI_batch and Additional file 2: Table S1 for additional evaluation metrics

From: One Cell At a Time (OCAT): a unified framework to integrate and analyze single-cell RNA-seq data