Table 2 Important features per descriptor. For each descriptor, we give the value of the highest influence and the number of features with this influence. We present the smallest of these features and another example
From: OWSum: algorithmic odor prediction and insight into structure-odor relationships
Descriptor | Floral | Perfumery | Woody, resinous | Medicinal | Sickening | Fruity, other than citrus |
|---|---|---|---|---|---|---|
Highest influence value [× 10–6] | 17 | 14 | 13 | 77 | 354 | 110 |
Number of features with this influence | 57 | 57 | 101 | 124 | 1 | 13 |
Smallest characteristic feature |
|
|
|
|
|
|
SMARTS of this feature | [#6](-[#6])(-[#6]) = [#6] | [#6](-[#6])(-[#6]) = [#6] | [#6](-[#6](-[#6]-[#6])-[#6]-[#6])-[#6] | [#6]:[#6]-[#8] | [OX2H1][CX3](= [OX1])[#6] | [#6](-[#6](-[#8]-[# 6]-[#6]) = [#8])-[#6] |
Smallest feature with another characteristic for this influence or second example |
|
|
|
| - |
|
SMARTS of this feature | [#6]-[#6] = [#6]-[#6] | [#6](-[#6]) = [#6]-[#6] | [#6](-[#6])-[#6](-[#6]-[#6]-[#6]-[#6])-[#6](-[#6])-[#6] | [#6]:1(:[#6]:[#6]:[#6]:[#6]:[#6]:1)-[#8] | - | [#8] = [#6](-[#6]-[#6])-[#8]-[#6]-[#6] |
