Table 2. Pearson’s Correlation for Few-Shot Learning Methods and Tanimoto Similarity on the BindingDB Test Set^a.

# context compounds	1	2	4	8
Tanimoto similarity	–0.06	0.11	0.17	0.23
MolBERT + ANP	0.10	0.11	0.11	0.14
NGGP	0.05	0.11	0.15	0.20
MetaNet	–0.02	–0.01	0.04	0.06
Meta-MGNN	0.20	0.20	0.21	0.23
MAML	0.22	0.22	0.23	0.24
MetaDTA	0.24	0.24	0.25	0.25
FS-CAP	0.27	0.28	0.29	0.32

^aResults are reported as the mean across all BindingDB targets of the Pearson’s r between predicted and ground-truth pActivity. We ran three replicate training runs with different random number seeds for the three best performing methods (MAML, MetaDTA, FS-CAP) and obtained standard deviations of at most 0.01.