The document contains chemical structures and mathematical equations. It discusses machine learning methods for predicting molecular properties from graph-based representations of molecules using node and edge descriptors without quantum chemistry calculations. Random forest, support vector machines, and other models are mentioned for making predictions from molecular graph encodings.
6. N
NH
OO
HH
H
H H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
O
O
O
O
O
O
Cl
H
H
H
H
H
HH
H
H
H
H
H
H
H
H
H
H
Br
Br O P
O
O Br
Br
O
Br
Br
H
H
H
H
H
H
H
H
H
H
HH
H
HH
N
S
N
N
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
O
N
O
O
H
H
H
O
O
H
H
N
O
O
Cl
ClCl
H
H
H
H
H
H H
N
O
O
H
H
H
H
H
H
H H
H
N
O
O
H
H
H
H
H
H
H
N
H
N
O
O
N
O
O
H
H
H
H
H
H
H
H
N
CH3
O
O
H
N Cl
Cl
Cl
Cl
Cl
H3C
O O
O
O
O
O
H3C
CH3
CH2
O
HN
O
O
NH
CH3
HO
OH
CH3
N
O
O
CH3
N
N
H
N
H
H3C
N
H3C
H3C
NH
O
N
O
NO
CH3
O N
NH2
O
CH3
Br
CH3
N
H3C
H
NS
N
O
CH3
N
OH
CH3
CH3N
N
N
CH3H3C
H2N NH2
H
OH
O
HO
CH3
H
H
O
CH3
H
O
OH3C HH
H
O
H3C
S
CH3
O
H
H
O
CH3
CH3
OO
HO
H3CH
HO
F
H
O
H3C
NH2
O
N
HO
HO
O
H
H
O
O
OH3C
O
O
O
CH3
O
CH3
HO
CH3
H
O
O
CH3
H
H
N
H
N O
H3C
O
O
O
38. • the number of immediate neighbors who are
“heavy” (non-hydrogen) atoms
• the valence minus the number of hydrogens
• the atomic number
• the atomic mass
• the atomic charge
• the number of attached hydrogens
• whether the atom is contained in at least one ring
• hydrogen-bond acceptor or not?
• hydrogen-bond donor or not?
• negatively ionizable or not?
• positively ionizable or not?
• aromatic or not?
• halogen or not?
Rogers+, Extended-Connectivity Fingerprints. J. Chem. Inf. Model., 2010, 50 (5), pp 742–754
Faber+, Prediction Errors of Molecular Machine Learning Models Lower than Hybrid DFT Error. J. Chem. Theory Comput., 2017, 13 (11), pp 5255–5264