Privacy-preserving search for chemical compound databases

BMC Bioinformatics

Table 1 The experimental success ratios of the user's guess based on the server's return and the prior distribution of true value (n = 813,

	n= 831	n= 831 × 10¹	n= 831 × 10²	n= 831 × 10³	n= 831 × 10⁴	Ideal value
w ^{ChEMBL−177159}	0.03552	0.01738	0.01101	0.01009	0.00977	0.00981
w ^{ChEMBL−265935}	0.02991	0.01337	0.00903	0.00798	0.00784	0.00807
w ^rand	0.00914	0.0041	0.00309	0.00279	0.00305	0.00289

${\bar{TI}}_{1, 1, 0.8} (k = 831)$ is assumed and results are calculated for five different numbers of dummies (n = 831, 831 × 10¹, 831 × 10², 831 × 103, 831 × 10⁴) are used for three different distributions: w^{ChEMBL−177159} and w^{ChEMBL−265935} are actual distributions of ${\bar{TI}}_{1, 1, 0.8}$ on ChEMBL obtained by querying two randomly selected fingerprints from ChEMBL, w^rand is obtained by randomly selecting a value from 1, ..., k for m = 5 × 831 times and dividing each observed frequency by m.

ISSN: 1471-2105