Skip to main content
. 2021 Nov 24;36(5):391–404. doi: 10.1007/s10822-021-00421-6

Fig. 2.

Fig. 2

Analysis of chronological split molecular similarity statistics across 788 protein targets. Mean values over each distribution are written in parentheses. (a) Comparison between mean test-test (0.73), train-train (0.74), and test-train (0.53) molecular similarities. (b) Comparison between training set and test set time slice (validation—0.60, test-early—0.53, and test-late—0.47) molecular similarities