Table 1:

iSIM JT, iSIM-JT, and Pairwise $\sigma$ for sampled molecules (10%) from ChEMBL33 natural products (whole dataset: n = 64086, iSIM = 0.2745, iSIM-$\sigma =0.1302$) using different sampling methods. Molecules were represented with RDKit fingerprints (2048 bits).

Sampling method	iSIM	iSIM $\mathit{\sigma }$	Pairwise $\mathit{\sigma }$
Medoid	0.6031	0.0747	0.0763
Outlier	0.0856	0.0687	0.0807
Extremes	0.2800	0.2742	0.2748
Quota	0.3113	0.1913	0.1890
Stratified	0.2745	0.1329	0.1302
MaxMin	0.1764	0.0936	0.0900