Skip to main content
Fig. 5 | BMC Bioinformatics

Fig. 5

From: Automated annotation of rare-cell types from single-cell RNA-sequencing data through synthetic oversampling

Fig. 5

Comparison of the baseline classification and sc-SynO visualized as mean outcome for the used quality parameter: F1-Score (grey), Precision (blue), and Recall (purple). Detailed results can be obtained in Table 3. We observe that in every case, sc-SynO improves the recall compared to the Baseline model (see dotted boxes in the figure). This ensures that oversampling with sc-SynO improves the detection rate of rare-cell types (recall of the classification). Five-fold cross validation is repeated five times with five different shuffles of the dataset, which is shown as error bars to represent the variance of the performance measures over five readings

Back to article page