A Genetic Algorithm Based Support Vector Machine Model for Blood-Brain Barrier Penetration Prediction

Table 4

The most frequently used features for all 6-feature models^a.

Number	Feature name	Occurrence; (50 models)	Meaning
11	AbsCarboxy	36	Indicator for carboxylic acid^†
268	ESP-FHASA_Fractional_HASA_(HASA/TMSA)_Quantum-Chemical_PC	14	H-acceptor surface area/total molecular surface area^#
101	Topographic_electronic_index_(all_bonds)_Zefirov’s_PC	12	Topological electronic index for all bonded pairs of atoms
8	M_PSA_7.4	11	PSA at pH 7.4
267	ESP-HASA_H-acceptors_surface_area_Quantum-Chemical_PC	10	H-acceptor surface area^#
5	delta_	9	(pH 6.5) − (pH 7.4)
7	M_PSA_7.0	9	PSA at pH 7.0
138	HA_dependent_HDCA-2_[Zefirov’s_PC]	9	H-donors charged surface area^#
6	M_PSA_6.5	8	PSA at pH 6.5
1	M_	7

Rows with the same symbol could be categorized into the same group.
^bTopological electronic index is a feature to characterize the distribution of molecular charge: , where is net charge on th atom and is the distance between two bonded atoms.
^c7.4 is the pH in blood.
^d6.5 is the pH in intestine.
^e is the ratio of the sum of the concentrations of all species of a compound in octanol to the sum of the concentrations of all species of the compound in water. For neutral compounds, is equal to .