One molecular fingerprint to rule them all: drugs, biomolecules, and the metabolome

Table 4 Nearest neighbor analysis of the human metabolome database

HMBD subset	All	OH = 0	OH = 1	1 < OH ≤ 4	OH > 4
All	96,456	33,721	10,663	41,493	10,579
JD (MAP4-1024) = 0	0	0	0	0	0
JD (AP) = 0	1677	13	35	1611	18
JD (TT) = 0	68,623	27,897	5782	32,909	2035
JD (MHFP6-1024) = 0	69,972	28,502	6215	33,359	1996
JD (ECFP4-1024) = 0	70,329	28,561	6243	33,294	2231

Subsets of the Human Metabolome 4.0 Database according to the number of hydroxyl groups per molecule separating lipids (OH = 0, 1) from carbohydrate derivatives (OH > 4). For each subset (column), the number of molecules is indicated in total (All, line 2) and counting those with an indistinguishable nearest neighbor (Jaccard Distance JD = 0) according to the indicated fingerprint (line 3–7). Molecules were considered after removing stereochemical information

ISSN: 1758-2946