dataset and 1398 Mtb inactive compounds reported in ChEMBL database
(molecular weight ranging from 180 to 400, number of heavy atoms ranging from
12 to 27, similar to SAM/SAHC and the 23 inhibitors) were used as the negative
dataset. Structures of these molecules were energy minimized and five lowest
energy conformers were chosen for each of them. All these conformations were
mapped to the 41 e-pharmacophore models using the ‘advanced pharmacophore
screening’ option of Phase. Fast conformational sampling was used during pharmacophore screen, excluding molecules with >15 rotatable bonds. Molecules,
which could be mapped to at least four pharmacophoric sites of each model were
screened and among several conformers of a molecule the one with the best fitness
score (S) given by the following equation [46] was retained for each compound. S is
a measure of volume overlap and extent of match of chemical nature and directionalities of the pharmacophoric features with the corresponding complementary
features of the molecules.
S ¼ W site 1 À S align =C align
À
Á þ W vec S vec þ W vol S vol þ W ivol S ivol
where W site ¼ ð1 À S align =C align Þ, S align = alignment score, C align = alignment cutoff, S vec = vector score, W vec = weight of vector score, S vol ðV common =V total Þ =
volume score, W vol = weight of volume score, S ivol = included volume score.
Detailed explanations of the components of the fitness score are given in reference
47. Volumes were computed using van der Waals models of all atoms except
non-polar hydrogens, and W ivol is the weight of volume score. C align , W site , W vec ,
W vol and W ivol are user-adjustable parameters, with default values of 1.20, 1.00,
1.00, 1.00 and 0.0, respectively.
Analysis of the hits obtained from these pharmacophore screening showed that
most of the models developed from the CmaA1 complexes obtained from the MD
trajectories were able to screen up to 17 reported inhibitors (out of 23), while the
model developed from the crystal structure could screen only one inhibitor.
Fig. 3 Generation of dynamics-based e-pharmacophore models from the MD trajectory. The
associated active site residues’ interactions have been shown. The colour representations for the
features are same as Fig. 1
42
C. Choudhury and G. Narahari Sastry
(molecular weight ranging from 180 to 400, number of heavy atoms ranging from
12 to 27, similar to SAM/SAHC and the 23 inhibitors) were used as the negative
dataset. Structures of these molecules were energy minimized and five lowest
energy conformers were chosen for each of them. All these conformations were
mapped to the 41 e-pharmacophore models using the ‘advanced pharmacophore
screening’ option of Phase. Fast conformational sampling was used during pharmacophore screen, excluding molecules with >15 rotatable bonds. Molecules,
which could be mapped to at least four pharmacophoric sites of each model were
screened and among several conformers of a molecule the one with the best fitness
score (S) given by the following equation [46] was retained for each compound. S is
a measure of volume overlap and extent of match of chemical nature and directionalities of the pharmacophoric features with the corresponding complementary
features of the molecules.
S ¼ W site 1 À S align =C align
À
Á þ W vec S vec þ W vol S vol þ W ivol S ivol
where W site ¼ ð1 À S align =C align Þ, S align = alignment score, C align = alignment cutoff, S vec = vector score, W vec = weight of vector score, S vol ðV common =V total Þ =
volume score, W vol = weight of volume score, S ivol = included volume score.
Detailed explanations of the components of the fitness score are given in reference
47. Volumes were computed using van der Waals models of all atoms except
non-polar hydrogens, and W ivol is the weight of volume score. C align , W site , W vec ,
W vol and W ivol are user-adjustable parameters, with default values of 1.20, 1.00,
1.00, 1.00 and 0.0, respectively.
Analysis of the hits obtained from these pharmacophore screening showed that
most of the models developed from the CmaA1 complexes obtained from the MD
trajectories were able to screen up to 17 reported inhibitors (out of 23), while the
model developed from the crystal structure could screen only one inhibitor.
Fig. 3 Generation of dynamics-based e-pharmacophore models from the MD trajectory. The
associated active site residues’ interactions have been shown. The colour representations for the
features are same as Fig. 1
42
C. Choudhury and G. Narahari Sastry