Skip to main content
. 2022 Oct 28;12:18200. doi: 10.1038/s41598-022-23014-1

Figure 4.

Figure 4

The prediction performance of the final DeepLPI model on BindingDB dataset. (a) The ROC curve and the determined optimal threshold. (b) Confusion matrix based on the optimal threshold. (c–e) Confusion matrix and performance metrics on the three “unseen” drug/protein testsets: (c) Molecule unseen, (d) Protein unseen and (e) None seen.