Skip to main content

Advertisement

Table 2 Recovery of fragrance molecule families from various databases

From: Expanding the fragrance chemical space for virtual screening

Fragrance Cpds nr. HAC av. FragranceDB recov. at 10% PubChem.FL recov. at 1% ChEMBL.FL recov. at 1% ZINC.FL recov. at 1% GDB-13.FL recov. at 0.1%
Vegetable 10 7.20 45/0/22/45 56/0/44/11 45/0/11/0 33/0/22/0 78/22/67/56
Fishy 11 8.64 40/20/40/0 40/30/40/0 50/20/20/0 10/20/40/0 67/44/78/33
Chemical 23 8.87 14/14/9/9 14/18/9/0 5/5/9/0 5/9/9/0 37/37/63/21
Ethereal 14 8.93 46/46/23/8 36/62/23/8 46/54/15/8 23/46/15/8 55/82/55/45
Medicinal 12 9.58 55/64/55/9 55/64/55/9 55/46/37/9 55/55/36/9 67/89/89/56
Nutty 28 10.14 37/30/4/15 33/37/4/4 22/19/9/4 19/19/4/4 42/54/13/21
Fatty 42 10.36 17/22/15/12 10/27/20/7 17/17/5/7 7/22/5/2 33/45/48/3
Smoky 12 11.42 18/18/36/9 18/18/27/8 9/9/18/0 9/9/18/0 -
Fruity 122 11.56 23/23/5/16 17/33/8/2 19/22/1/8 11/21/2/2 35/49/36/0
Minty 13 11.92 58/8/50/33 42/0/42/8 42/0/34/8 42/0/42/8 44/0/22/22
Citrus 35 12.06 29/15/12/18 9/18/18/0 36/15/12/0 9/15/18/0 9/30/43/13
Balsamic 64 12.25 30/6/5/13 19/6/8/2 14/2/2/2 5/5/0/2 39/10/29/0
Floral 69 12.81 22/0/16/21 7/0/12/6 9/0/6/6 6/0/6/6 18/0/43/7
Herbaceous 13 12.92 33/17/8/17 8/0/0/8 8/0/0/8 8/0/0/8 -
Waxy 11 14.18 60/40/40/30 30/40/90/10 50/40/40/10 30/40/70/10 -
Average 32 10.86 35/22/23/17 26/24/27/5 29/17/14/5 18/17/19/4 44/39/49/23
No. of best scores per series 12/5/2/1 5/6/6/1 11/3/2/1 7/7/7/2 3/4/6/0
  1. For each database the % actives found is given for the indicated % database screened by sorting with MQN/Sfp/ECfp4/MW similarity to the most average molecule in the set. The highest value in each entry is highlighted in bold. Fragrance families were collected from the Superscent database website. Compounds appearing in more than 5 different families and those not following FL criteria were removed. Data was not computed for GDB-13.FL if the families were smaller than 10 compounds after removal of HAC > 13 compounds. The city-block distance was used as similarity measure (results were comparable using Tanimoto).