Skip to main content
. 2023 Apr 1;14:1836. doi: 10.1038/s41467-023-37478-w

Fig. 3. Detailed results on the accuracy performances of the DR methods.

Fig. 3

a The accuracy performances of the DR methods on all real CyTOF datasets, grouped by the 11 cohorts. Orange color indicates better performance, while blue color indicates worse performance. All submetrics of the accuracy criterion were averaged. “*” indicates that forced 10% downsampling was performed for training in the case that the methods aborted because of sample size. “scvis” is always performed on 10% down-samples per authors’ recommendation. Small blue dot indicates error. b The accuracy performances of the DR methods for each accuracy category. Results for all real CyTOF datasets were averaged, and the results for each detailed sub-metric of accuracy was shown. Orange color indicates better performance, while blue color indicates worse performance. c The accuracy performances of the DR methods on the simulated CyTOF data. The performances were visualized as a function of the characteristics of the CyTOF datasets. Orange color indicates better performance, while blue color indicates worse performance. All submetrics of the accuracy criterion were averaged. “*” Indicates that 10% downsampling was performed. Source data are provided as a Source Data file.