Skip to main content
. 2021 Oct 28;38(3):597–603. doi: 10.1093/bioinformatics/btab746

Fig. 3.

Fig. 3.

Dependency of performance on sequencing depth for the Ser dataset. (Left) ROC AUC in function of coverage (# reads) of the label. On top of the plot the percentage of labels belonging to each bin is shown. (Middle) ROC AUC in function of local sparsity (defined as the percentage of unobserved entries in the local window used for prediction). (Right) ROC AUC in function of both factors for CpG Transformer. The biggest gradient in performance is observed for the local sparsity direction