Fig. 2: Comparison of integration performance over partially overlapping datasets by different methods.

a UMAP embeddings of the integration results by the indicated methods based on the liver dataset. b Over-correction score of different methods based on the indicated benchmark datasets. c Over-correction score of the indicated methods based on the simulated datasets, with decreased numbers of common cell-types (obtained by down-sampling the pancreas and PBMC dataset). d UMAP embeddings of the integration results by the indicated methods based on the simulated pancreas datasets with different numbers of specified common cell-types. Over-corrections are highlighted with red circles and error-splittings with purple circles, respectively.