PubChem structure–activity relationship (SAR) clusters

Table 2 Average (\(\bar{x}\)) and standard deviation (s) of the similarity scores between 10,000 randomly-selected biologically-tested compounds (from Ref. [22, 23]), and the dissimilarity threshold (d ^thresh) used in the present study to generate the structure–activity relationship (SAR) clusters

Similarity measures	\(\bar{x}\)	s	\(\bar{x}\) + 2s	d ^thresh
2-D	0.4229	0.1326	0.6881	0.3119
3-D (N _max = 1)
ST ^ST-opt	0.5438	0.0986	0.7410	–
ComboT ^ST-opt	0.6161	0.1276	0.8713	–
CT ^CT-opt	0.1807	0.0609	0.3024	–
ComboT ^CT-opt	0.5859	0.1440	0.8738	–
3-D (N _max = 10)
ST ^ST-opt	0.6464	0.1017	0.8498	0.1502
ComboT ^ST-opt	0.7682	0.1337	1.0356	0.4822
CT ^CT-opt	0.2485	0.0706	0.3898	0.6102
ComboT ^CT-opt	0.7733	0.1386	1.0505	0.4748

N _max is the maximum number of diverse conformers considered per compound for the 3-D similarity computation. The d ^thresh value for each of the five similarity measures were determined by subtracting its (\(\bar{x}\) + 2s) value from unity (after normalization to one for ComboT ^ST-opt and ComboT ^CT-opt). The statistical parameters for N _max = 10 were used to determine the d ^thresh value for the 3-D similarity measures.

ISSN: 1758-2946