Pharmacology & Pharmacy, 2011, 2, 271-281
doi:10.4236/pp.2011.24035 Published Online October 2011 (http://www.SciRP.org/journal/pp)
Copyright © 2011 SciRes. PP
271
Modeling of the Interaction of Flavanoids with
GABA (A) Receptor Using PRECLAV
(Property-Evaluation by Class Variables)
Vijay K. Agrawal1, Basheerulla Shaik1, Padmakar V. Khadikar1, Shalini Singh2
1Department of Applied Sciences, National Institute of Technical Teachers’ Training & Research, Bhopal, India; 2Depatment of
Chemistry, Bareily College, Bareily, India.
Email: basheerulla.81@gmail.com, pvkhadikar@rediffmail.com, shalinisingh_15@yahoo.com, apsvka@yahoo.co.in
Received April 28th, 2011; revised August 17th, 2011; accepted September 20th, 2011.
ABSTRACT
Quantitative Structure-Activity Relationship (2D-QSAR) models for binding affinity constants (log Ki) of 78 flavanoid
ligands towards the benzodiazepine site of GABA (A) receptor complex were estimated using the PRECLAV (Prop-
erty-Evaluation by Class Variables) program. The best MLR equation with nine PRECLAV descriptors has R2 = 0.843
and C = 0.782. Attempt is also made for obtaining 2D-QSAR model using NCSS software. The comparison of the
results indicated that the PRECLAV method is very efficient in detecting structure-activity correlation with good pre-
dictive power.
2
R
Keywords: PRECLAV, NCSS, Regression Analysis, Cross-Validation, GABA, Flavanoids
1. Introduction
During the last two decades quantitative structure-activ-
ity relationship (2D-QSAR) models have gained exten-
sive recognition in drug design [1]. The widespread use
of 2D-QSAR models come from the development of
novel structural descriptors and statistical equations re-
lating activity with chemical structure. The main hy-
pothesis in the 2D-QSAR approach is that all biological
activity of a chemical substance is statistically related to
its molecular structure. The PRECLAV program uses the
atom in the common skelton to compute bond and field
(grid) descriptors [2,3]. The PRECLAV program com-
putes five classes of structural descriptors: Constitutional,
topological indices, molecular graph invariants, geomet-
rical, quantum bond indices and field (grid) descriptors
[2-4].
All molecules are aligned by superimposing the com-
mon atom before generating the multiple linear regres-
sion models; PRECLAV makes a descriptor selection by
discarding those descriptors that are poorly correlated
with the investigated activity.
During last decade more than 400 chemically unique
flavonoids (phenyl-benzopyrans) have been isolated from
vascular plants and many of them are used as tranquiliz-
ers in folkloric medicine. Such type of compounds are
important constituents of the human diet, being derived
largely from fruits and vegetables, nuts, seeds, stems and
flowers and thus constitute one of the important classes
of the metabolites. Some of the compounds from fla-
vones family exhibit a potent in vivo anxiolitic activity,
and do not involve unwanted side effect. As a result of
this several attempts have been made to generate syn-
thetic flavones derivatives with higher affinities for the
GABA (A) receptor [5-10]. Subsequently, attempts were
also made to establish quantitative structure-activity rela-
tionship so as to establish a 2D-QSAR model for inhibi-
tion of GABA (A) receptor that could serve as a guide
for the rational design of further potent and selective in-
hibition having the flavones backbone [11-13]. One such
attempt was recently made by Duchowiz and co workers
[14-16]. They have proposed the best linear model for a
set of 70 flavones and found that the best model involves
four correlating descriptors with statistical quality given
by R2 = 0.7174, Se = 0.580, = 0.6757, SLOO =
0.622.
2
LOO
R
It was observed that out of several available software’s
such as COMFA [17], CORBA [18], OASIS [19],
CODESSA [20], TSAR [21], PRECLAV [2,3], etc. The
PRECLAV software is very efficient in detecting struc-
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
272
(Property-Evaluation by Class Variables)
ture-activity relationship with good predictive power.
This has prompted us to use PRECLAV program for in-
vestigating GABA (A) receptor binding and to compare
the findings with those obtained using NCSS software.
2. Database and Modeling
The data base used as input by PRECLAV consists of 78
flavoniods presented in Table 1 together with their log
Ki(μM) values [14]. The chemical structures were gener-
ated with Hyper Chem. [22], geometry optimization was
performed with MOPAC [23] and the QSAR models
were computed with PRECLAV [2,3]. MOPAC 7 output
files are used by PRECLAV [2,3] program to compute
PRECLAV descriptors for generating multiple linear
regression models. Before such generation of the models
PRECLAV software makes a descriptor selection by
discarding those descriptors that are poorly correlated
with the investigated activity. The following descriptors
were generated in the present case:
3. Notations of the Structural Descriptors
Generated by PRECLAV
1) MATS2p: Moran autocorrelation-lag 2/weighted by
atomic polarizabilities (2-D autocorrelation indices) Dragon
Descriptor.
2) OXX: presence of Oxygen. Maximum charge for O
atom (at parabolic region) PRECLAV Descriptor.
3) NGS: area of negative charged surface/molecular
surface area ratio (at parabolic region) PRECLAV De-
scriptor.
4) HBA: Capability to form.
5) Hydrogen bonded (function No. 1) (at parabolic re-
gion) PRECLAV Descriptor.
6) VLS: volume of circumscribed sphere (at parabolic
region) PRECLAV Descriptor.
7) B05[O-B]: presence/absence of [O-B] at topological
distance 05(2D binary fingerprint) Dragon Descriptor.
8) GVWAI-50: Ghose-Viswanadhan-Wendoloski drug-
like index at 50% (molecular properties) Dragon De-
scriptor.
9) B08[C-O]: presence/absence of [C-O] at topological
distance 08. (2D b inaryfingerprint) Dragon Descriptor.
10) HTm: H total index/weighted by atomic masses
(GETAWAY descriptors).
These descriptors are chosen on the basis of their qual-
ity (Q) and were used to generate the best MLR (Multi-
linear regression) model.
Finally, the leave-one-out (LOO) cross-validation pro-
cedure is applied to each and every MLR equation in
order to estimate the prediction power of the proposed
QSAR equations. The predictive ability of a QSAR
equation is estimated with the LOO Pearson and Rank
(Kendall) correlation coefficients and . The
equation with the highest predictive power is considered
to be the one with the highest value for the product
× . This QSAR model can further be used to pre-
dict the activity of novel, not yet tested compounds
(Drugs).
2
cv
R2
Kendall
R
2
cv
R
2
Kendall
R
In the present study for modeling log Ki of 78 com-
pounds initially we have used 400 PRECLAV and 1457
DRAGON descriptors. The number of excluded near
constant descriptors being 89, while the number of sig-
nificant descriptors is 174. One by one outliers is re-
moved from calibration set so that final 2D-QSAR model
is obtained.
4. Results and Discussion
After computing the structural descriptors for the 78 fla-
vones (Table 1) PRECLAV performs the descriptors
solution and generation of best QSAR equation. Because
it is important to have a reference for the evaluation of
MLR model, we give here correlations prediction/prop-
erty of the aforementioned most valuable predictors
MATS2P, 0.444175: OXX, 0.2524; NGS, 0.1232; HBA,
0.1232; VLS, 0.112; BO5[O-Br], 0.1035; GVWAI-50,
0.0886; BO8[C-O], 0.0836; HTm, 0.0607.
During the PRECLAV MLR analysis, we observed
that the equation with highest value of the ×
is the 7-parametric models and that this model also has
the highest predictive power and is as follows.
2
cv
R2
Kendall
R
log Ki = –2.5590 – 14.7642 MATS2P + 0.8940 OXX
+ 0.5971 NGS – 1.0633 HBA + 1.0633 XNC
+ 0.0656 HTm + 4.3878 R2U.
N = 78, R2 = 0.7150, F = 24.447, = 0.6424,
= 0.6459, Rkcv = 0.6063, Q = 0.5879.
2
kendal
R
2
cv
R
A detailed regression analysis of this model using
PRECLAV software indicated that there are seven com-
pounds which are (67, 59, 40, 25, 38, 46, and 64) acting
as outliers. These compounds are, therefore, removed for
obtaining an appropriate model.
Also, note that in the above model the coefficient of
MATS2P, HBA, and GVWAI-50 are negative indicating
that log Ki increases with decrease in the magnitude of
these predictors. Both Pearson and Kendall LOO coeffi-
cients are high showing that the equation can provide
reliable predictions. Furthermore, despite the larger
number of structural descriptors in the above equation
there is no evidence of over fitting, as indicated by high
values of and .
2
cv
R2
Kendal
R
It is worth mentioning that one by one removal of
compounds 67, 59, 40, 25, 38, 36, and 64 acting as out-
liers resulted into both changes in the number of de-
scriptors as well as the regression quality. The statistical
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV 273
(Property-Evaluation by Class Variables)
Table 1. List of 78 flavoniods and their observed log Ki val-
ues.
O
O
Compd. No Compound Obs. log Ki
1 6-F-luoro-3’-methoxyflavone 0.398
2 6-Bromo-3’-mcthoxyflavone –0.215
3 6-Chloro-4’-methoxyflavonc 0.097
4 6-Bromo-4’-methoxyflavone 0.322
5 6-Chloro-2’-fluoroflavone –0.380
6 6-Bromo-2’-fluoroflavone –0.424
7 6,3’-Difluoroflavone –0.036
8 6-Chloro-3’-fluoroflavone 0.9.2
9 6-Bromo-3’-fluoroflavone –1.377
10 4’-Fluoroflavonc 0.556
11 6.4’-Difluoroflavone 0.398
12 6-Chloro-4’-fluoroflavone –0.742
13 3’-Chloroflavone –0.212
14 6,3’-Dichloroflavone –1.638
15 3’-Bromoflavone –0.384
16 6-Fluoro-3’-bromoflavone –0.627
17 6-Chloro-3’-bromoflavone –1.638
18 6,3’-Dibromollavone –1.721
19 6-Bromo-4’-nitroflavone –0.699
20 6-Bromoflavone –1.155
21 6-Chloroflavone –0.785
22 6-Nitroflavone –0.561
23 6-Methoxyllavone –0.066
24 6-Fluoroflavone 0.653
25 6-Bromo-3’-nitroflavone –3.000
26 6-Methyl-3’-nitroflavone –2.252
27 6-Chloro-3’-nitroflavone –2.097
28 6-3’-Dinitroflavone –1.585
29 6-Fluoro-3’-nitroflavone –0.745
30 3’-Nitroflavonc –0.545
31 6-Methyl-3’-bromoflavone –1.886
32 6-Nitro-3’-bromoflavone –1.602
33 6-Hydroxy-3’-bromoflavone 0.000
34 6-Methoxy-3’-bromoflavone 0.000
35 6-3’-Dimethylflavone –0.682
36 3’-Methylflavone 1.000
37 5,2’-Dihydroxy-6,7,8,6’-tetramethoxy
flavone –0.444
38 5,7,2’-Trihydroxy-6,8-dimethoxyflavone –2.215
39 2’-Hydroxy-a-napthoflavone –1.569
40 6.2’-Dihydroxyflavone –1.469
41 5,7,2’-Trihydroxy-b.-methoxyflavone –1.420
42 5,7,2’-Trihydroxyflavone –1.125
43 2’-Hydroxyflavone –0.678
44 5.7-Dihydroxy-6.8-dimethoxyflavone –0.699
45 7,2’-Dihydroxyflavone –0.252
46 5,7-Dihydroxy-6-methoxyflavone –0.051
47 5.7-Dihydroxy-8-methoxyflavone 0.182
48 6-F Hydroxyflavone 0.422
49 7-Hydroxyflavone 0.623
50 5,6,7-Trihydroxyflavone 0.747
51 6-Hydroxy-2’-mcthoxytlavone 0.976
52 2’-Methoxyflavone 1.508
53 2’-Amino-6-methoxyflavone 0.544
54 Flavone 0.000
55 5,7-Dihydroxyflavone 0.477
56 5,3’,4’-Trihydroxy-6,7-dimethoxyflavone 2.301
57 5,4’-Dihydroxy-6,7-dimcthoxyflavone 1.362
58 5,7,4’-Trihydroxy-6-mcthoxyflavone 0.000
59 5,7-Dihydroxy-2’-chloroflavone 0.903
60 5,7-Dihydroxy-2’-fluoroflavone 0.903
61 5,7-Dihydroxy-6,8-dibromoflavone –0.155
62 5,7,4’-Trihydroxyflavone 0.602
63 3,5,7,4’-Tetrahydroxytlavone 1.969
64 5-Hydroxy-7-methoxyllavone 1.699
65 5,7-Dihydroxy-6,8-diiodollavone 0.000
66 6-Fluoro,3’-hydroxyflavone 0.400
67 6-Chloro,3’-hydroxyflavone 0.070
68 6-Bromo,3’-hydroxyflavone 0.220
69 6-Bromo-2’-nitrotlavone 0.680
70 6-Nitro-4’-bromollavone 1.600
71 3’-Methoxyflavone 0.380
72 6-Chloro-3’-methoxyflavone 0.072
73 3’-Fluoroflavone 0.550
74 6-Bromo-4’-fluoroflavone 0.939
75 6-Fluoro-3’-chlorollavone 0.701
76 6-Bromo-3’-chlorollavone 1.770
77 6-Methylflavone 0.903
78 6-Bromo-3’-methylflavone 0.812
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
(Property-Evaluation by Class Variables)
Copyright © 2011 SciRes. PP
274
nine correlating parameters using variable selection ana-
lysis (Table 2). These parameters are the same as those
were used in the aforementioned PRECLAV QSAR mod-
eling. However, unlike PRECLAV, the NCSS programs
clearly demonstrate successive arrival of 9-para- metric
model. (Table 2). Among the regression results the best
one-, two-, three-, four-, five-, six-, seven-, eight- and nine-
parametric models were selected and are given in Table 3.
detail of this model is given below:
log Ki = –2.9314 – 16.6179 MATS2P + 0.8709O XX
+ 0.4905 NGS – 1.8923 HBA + 0.3671 VLS
– 0.34610.5135 BO5[O-Br]
– 0.3461 GVWAI-50 + 0.7031 BO8[C-O]
+ 0.0489 HTm
N = 71, R2 = 0.8098, F = 29.333, = 0.7095,
= 0.7494, Rkcv = 0.6732, Q = 0.6544
2
kendal
R
2
cv
RIn these models, the correlation coefficient, R2, is a
measure of the fit of the model. F, the Fisher test value,
reflects the ratio of the variance explained by the model
and the variance due to the error in the model. Higher
values of F-test indicate the significance of the model.
We observed that the quality and predictive power of
the earlier model is considerably improved after deletion
of outliers. Furthermore, the physical significance of the
involved parameters is the same as before.
A perusal of Tab le 3 shows that using NCSS software
statistically allowed models start pouring with two- and
higher-parametric modeling. The regression parameters
and quality of these models are given below:
We have also used PRECLAV descriptors for obtain-
ing the best 2D-QSAR model using NCSS software.
Variable selection for multiple regression analysis has
demonstrated the occurrence of best regression model with
Table 2. Variable selection for multiple regression using NCSS.
Model No R2 R
2-Change Names
1 0.4175 0.4175 MATS2p
2 0.5359 0.1183 MATS2p, OXX
3 0.5940 0.0581 MATS2p, OXX, HTm
4 0.6605 0.0665 MATS2p, OXX, HBA, HTm
5 0.7140 0.0535 MATS2p, OXX, NGS, HBA, HTm
6 0.7462 0.0323 MATS2p, OXX, NGS, HBA, B08[C-O], HTm
7 0.7714 0.0252 MATS2p, OXX, NGS, HBA, B05[O-Br], B08[C-O], HTm
8 0.7929 0.0215 MATS2p, OXX, NGS, HBA, VLS, B05[O-Br], B08[C-O], HTm
9 0.8098 0.0169 MATS2p, OXX, NGS, HBA, VLS, B05[O-Br], GVWAI-50, B08[C-O], HTm
Table 3. Quality of regression and predictive potential for one to nine variable models.
Model No R2 2
A
R CV F Q PRESS/SSY2
CV
R SPRESS PSE
1 0.4175 0.4091 –2.2701 49.464 –0.2846 1.3950 –0.3950 0.7260 0.7157
2 0.5359 0.5222 –2.0413 39.255 –0.3586 0.8661 0.1339 0.6528 0.6389
3 0.5940 0.5758 –1.9234 32.675 –0.4007 0.6835 0.3165 0.6151 0.5975
4 0.6605 0.6399 –1.7722 32.095 –0.4586 0.5141 0.4859 0.5668 0.5464
5 0.7140 0.6920 –1.6391 32.448 –0.5155 0.4006 0.5994 0.5242 0.5015
6 0.7462 0.7225 –1.5558 31.368 –0.5552 0.3401 0.6599 0.4976 0.4724
7 0.7714 0.7460 –1.4483 30.372 –0.6064 0.2963 0.7037 0.4760 0.4484
8 0.7929 0.7662 –1.4280 29.672 –0.6236 0.2612 0.7388 0.4567 0.4268
9 0.8098 0.7818 –1.3797 28.860 –0.6522 0.2349 0.7651 0.4412 0.4090
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV 275
(Property-Evaluation by Class Variables)
Two-variable model
log Ki = –0.6987 – 7.0909 (±1.1003) MATS2p
+ 0.6765 (± 0.1625) OXX
N = 71, R2 = 0.5339, = 0.5222, CV = –2.0413,
F = 39.255
2
A
R
The positive coefficient of the parameter OXX indi-
cates that presence of Oxygen Maximum charge for O
atom (at parabolic region) is favourable for the exhibition
of the activity.
Three-variable model
log Ki = –1.2772 – 9.4417 (±1.2848) MATS2p
+ 0.7375 (±0.1544) OXX
+ 0.0296 (±0.0095) HTm
N = 71, R2 = 0.5940, = 0.5758, CV= –1.9234,
F = 32.675
2
A
R
Here the coefficients of both the parameters OXX and
HTm are positive meaning thereby that presence of
Oxygen Maximum charge for O atom (at parabolic region)
as well as H total index/weighted by atomic masses are
favourable for the exhibition of the activity.
Four-variabl e mo de l
log Ki = –2.1460 – 13.1294 (±1.5666) MATS2p
+ 0.9464 (±0.1536) OXX
+ 0.0471 (±0.0101) HTm
– 1.1215 (±0.3120) HBA
N = 71, R2 = 0.6605, = 0.6399, CV = –1.7722,
F = 32.095
2
A
R
Here also the parameters OXX and HTm have positive
coefficients meaning thereby that presence of Oxygen
Maximum charge for O atom (at parabolic region) as well
as H total index/weighted by atomic masses are favour-
able for the exhibition of the activity
Five-variable model
log Ki = –2.0375 – 13.1086 (±1.4489) MATS2p
+ 0.9128 (±0.1424) OXX
+ 0.0499 (±0.0093) HTm
– 1.2511 (±0.291 HTm 0) HBA
+ 0.6149 (±0.1763) NGS
N = 71, R2 = 0.7140, = 0.6920, CV= –1.6391,
F = 32.448
2
A
R
In this model, in addition to the two parameters OXX
and HTm the third parameter NGS has positive coeffi-
cient. This means that in addition to presence of Oxygen
Maximum charge for O atom (at parabolic region) as well
as H total index/weighted by atomic masses, the area of
negative charged surface/molecular surface area ratio (at
parabolic region) is also favourable for the exhibition of
the activity.
Six-variable model
log Ki = –2.7781 – 14.5970 (±1.4709) MATS2p
+ 0.8088 (±0.1400) OXX
+ 0.0557 (±0.0091) HTm
– 1.7353 (±0.3242) HBA
+ 0.5118 (±0.1712) NGS
+ 0.4724 (±0.1656) B08[C-O]
N = 71, R2 = 0.7462, = 0.7225, CV= –1.5558,
F = 31.368
2
A
R
We observe that in this model, in addition to the afore-
mentioned three parameters a fourth parameter viz. B08
[C-O] has positive coefficient clearly meaning thereby that
presence of Oxygen Maximum charge for O atom (at
parabolic region) as well as H total index/weighted by
atomic masses, the area of negative charged surface/ mo-
lecular surface area ratio (at parabolic region), the pres-
ence/absence of C-O at topological distance 08. (2D bi-
nary fingerprint) also favours the exhibition of the activity.
Seven-variable model
log Ki = –3.0056 – 16.0272 (±1.5082) MATS2p
+ 0.8461 (±0.1347) OXX
+ 0.0509 (±0.0089) HTm
– 1.8204 (±0.3118) HBA
+ 0.5333 (±0.1640) NGS
+ 0.5286 (±0.1598) B08[C-O]
+ 0.4638 (±0.1761) BO5[O-Br]
N = 71, R2 = 0.7714, = 0.7460, CV= –1.4883,
F = 30.372
2
A
R
In this model, in addition to the positive coefficients of
the aforementioned four parameters, the fifth parameter
namely BO5[O-Br] also has positive coefficient. It means
that in addition to presence of Oxygen Maximum charge
for O atom (at parabolic region) as well as H total in-
dex/weighted by atomic masses, the area of negative
charged surface/molecular surface area ratio (at parabolic
region), the presence/absence of C-O at topological dis-
tance 08. (2D binary fingerprint), the presence/absence of
O-B at topological distance 05 (2D binary fingerprint)
also favours the exhibition of the activity.
Eight-variable model
log Ki = –2.7789 – 15.4338 (±1.4659) MATS2p
+ 0.8315 (±0.1294) OXX
+ 0.0505 (±0.0085) HTm
– 1.7761 (±0.2996) HBA
+ 0.5435 (±0.1574) NGS
+ 0.4955 (±0.1539) B08[C-O]
+ 0.4533 (±0.1690) BO5[O-Br]
+ 0.4045 (±0.1595) VLS
N = 71, R2 = 0.7929, = 0.7662, CV= –1.4280,
F = 29.672
2
A
R
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
(Property-Evaluation by Class Variables)
Copyright © 2011 SciRes. PP
276
Here, we observe that in addition to the positive coef-
ficients of the above mentioned five parameters, the six
parameter namely VLS also has positive coefficient. This
clearly means that in addition to the presence of Oxygen
Maximum charge for O atom (at parabolic region) as well
as H total index/weighted by atomic masses, the area of
negative charged surface/molecular surface area ratio (at
parabolic region), the presence/absence of [C-O] at topo-
logical distance 08. (2D binary fingerprint), the presence/
absence of [O-B] at topological distance 05 (2D binary
fingerprint), volume of circumscribed sphere (at parabolic
region) also favours the exhibition of the activity.
Nine-variabl e mo de l
log Ki = –2.9314 – 16.6178 (±1.5048) MATS2p
+ 0.8709 (±0.1261) OXX
+ 0.0489 (±0.0083) HTm
– 1.8923 (±0.2938) HBA
+ 0.4905 (±0.1538) NGS
+ 0.7031 (±0.1734) B08[C-O]
+ 0.5135 (±0.1653) BO5[O-Br]
+ 0.3672 (±0.1549) VLS
– 0.3461 (±0.1486) GVWAI-50
N = 71, R2 = 0.8098, = 0.7818, CV= –1.3797,
F = 28.860
2
A
R
We observe that in this 9-parametric model the afore-
mentioned six correlating parameters have positive coef-
ficients .This means that their physical significance in
this model is the same as that of the 8-parametric model
discussed above.
The aforementioned 9-variable model is, therefore, the
most appropriate model and is subjected to Ridge regres-
sion [25] for investigating the existence or otherwise of
any co-linearity defect. The Ridge parameters, namely
VIF (variance inflation factor), T (Tolerance), CN (Con-
dition number), have been calculated and presented in
Table 4. We observed that VIF (variance inflation fac-
tor)values are much smaller than the allowed range of 10.
Also, that condition number for for the correlating pa-
rameters all are much lower than 100 and the tolerance
are <1. These observations therefore, suggest that no co-
linearity defect is present in the proposed model.
Relative performance of PRECLAV and NCSS soft-
ware
In order to further investigate the relative performance
of both PRECLAV as well as NCSS software we have
calculated (estimated) log Ki values for the 9-parametric
models using both softwares and compared them with the
experimental values of log Ki (Table 5). This is demon-
strated in Figures 1 and 2 indicating that quality of the
model obtain from both PRECLAV and NCSS software
is more or less same. log Ki values are much closer to the
experimental values in case of PRECLAV software.
From the study made herein we cannot definitely say as
to which software is superior. Both have their own merits
and demerits. However, the number of good points are
more in PRECLAV software as compared to NCSS
software. From the results obtained we conclude that
there are some good or bad points in both the software
and that overall PRECLAV software yields better statis-
tics compared to NCSS software. The comparison of the
performance of this software is demonstrated as below:
Comparison of results obtained using PRECLAV and NCSS softw are .
PRECLAV NCSS
1) Overall the best model is proposed 1) Recommends obtaining of the best model in succession which need to
be confirmed by their means
2) Predicts and removes the outliers one by one during the regression so
that the final model does not have any outlier 2) This is not possible
3) Performance cross-validation 3) Not possible
4) Selects most significant descriptors by quality 4) Selection of descriptors is not based on quality
5) Most valuable descriptors set is generated 5) Not possible
6) Correlation of predictor/activity is possible 6) Not possible
7) Yields inter correlation of predictors 7) Not possible
8) Makes estimated and observed values in calibration set 8) Yes it is also possible in NCSS
9) Analysis virtual fragments is possible 9) Not possible
10) Standard deviation of coefficients are not measured 10) We can estimate standard error of the coefficient of the correlating
parameters
11) Ridge statistics is not possible 11) We can make ridge analysis and then investigate co- linearity defect
12) Estimate qulity Q of the model 12) Not possible
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV 277
(Property-Evaluation by Class Variables)
It is worth mentioning that one of the important fea-
tures of PRECLAV software is the analysis of virtual
fragments. The software has indicated that for the set of
78 molecules analyzed here 30 virtual fragments are pre-
sent out of which 9 fragments are significant. These most
significant virtual fragments by correlation of “The Mass
percent” and “Property values” are given in Table 6.
This Table 6 demonstrates that large mass percent of CO,
C9H4O4, C6H5O7, C6H4O, and C8H5O4 increases logKi
values while the large mass percent at C8H4O, NO2, Br
and C6H4 decreases the log Ki values. These observations,
therefore, be taken care of while synthesizing new fla-
vones with better log Ki values.
In order to confirm our findings we have compared the
estimated values of the activities (log Ki) with the ex-
perimental ones (log Ki) (Table 5). This has further been
demonstrated in Figures 1 and 2. Also, we have obtained
Ridge traces as shown in Figures 3 and 4. From Figures
1 and 2 as well as Table 5, we observed that the esti-
mated activities (log Ki) are very close to the experimen-
tal activities (log Ki). Similarly, Figures 3 and 4 indi-
cates absence of any co-linearity defect.
Table 4. Ridge parameters for the most significant model.
Variable VIF T λi CN
MATS2p 4.4014 0.2272 3.367143 1.00
OXX 1.4179 0.7053 1.446224 2.33
NGS 1.1562 0.8649 0.957829 3.52
HBA 2.9861 0.3349 0.919456 3.66
VLS 1.0819 0.9243 0.771579 4.36
B05[O-Br]1.9536 0.5119 0.643925 5.23
GVWAI-501.8986 0.5267 0.460259 7.32
B08[CO] 2.5441 0.3931 0.316425 10.64
HTm 2.3977 0.4171 0.117160 28.74
Table 5. Comparison of log Ki values estimated using PRECLAV and NCSS Software.
Using PRECLAV Using NCSS
Compd. No Obs. log Ki Est. log Ki Residual Est. log Ki Residual
1 0.398 0.501 –0.103 0.501 –0.103
2 –0.215 –0.373 0.158 –0.373 0.158
3 0.097 0.228 –0.131 0.228 –0.131
4 0.322 0.084 0.238 0.084 0.238
5 –0.380 –0.898 0.518 –0.898 0.518
6 –0.424 0.067 –0.491 0.067 –0.491
7 –0.036 0.067 –0.103 0.067 –0.103
8 –0.932 –0.855 –0.077 –0.855 –0.077
9 –1.377 –1.013 –0.364 –1.013 –0.364
10 0.556 –0.118 0.674 –0.118 0.674
11 0.398 –0.042 0.44 –0.042 0.440
12 –0.742 –0.743 0.001 –0.743 0.001
13 –0.212 –0.76 0.548 –0.760 0.548
14 –1.638 –1.446 –0.192 –1.446 –0.192
15 –0.384 –0.641 0.257 –0.641 0.257
16 –0.627 –0.267 –0.36 –0.267 –0.360
17 –1.638 –1.302 –0.336 –1.302 –0.336
18 –1.721 –1.208 –0.513 –1.208 –0.513
19 –0.699 –0.828 0.129 –0.828 0.129
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
278
(Property-Evaluation by Class Variables)
20 –1.155 –1.285 0.13 –1.285 0.130
21 –0.785 –1.056 0.271 –1.056 0.271
22 –0.561 –0.992 0.431 –0.992 0.431
23 –0.066 0.535 –0.601 0.535 –0.601
24 0.653 0.256 0.397 0.256 0.397
25 Outlier
26 –2.252 –1.941 –0.311 –1.941 –0.311
27 –2.097 –1.751 –0.346 –1.751 –0.346
28 –1.585 –1.868 0.283 –1.868 0.283
29 –0.745 –0.463 –0.282 –0.463 –0.282
30 –0.545 –1.07 0.525 –1.070 0.525
31 –1.886 –1.285 –0.601 –1.285 –0.601
32 –1.602 –0.92 –0.682 –0.920 –0.682
33 0.000 –0.015 0.015 –0.015 0.015
34 0.000 –0.414 0.414 –0.414 0.414
35 –0.682 –0.631 –0.051 –0.631 –0.051
36 Outlier
37 –0.444 –0.485 0.485 –0.485 0.041
38 Outlier
39 –1.569 –1.114 –0.455 –1.114 –0.455
40 Outlier
41 –1.420 –0.757 –0.663 –0.757 –0.663
42 –1.125 –0.293 –0.832 –0.293 –0.832
43 –0.678 0.178 –0.856 0.178 –0.856
44 –0.699 –0.976 0.277 –0.976 0.277
45 –0.252 0.148 –0.4 0.148 –0.400
46 –0.051 0.131 –0.182 0.131 –0.182
47 0.182 0.004 0.178 0.004 0.178
48 0.422 0.948 –0.526 0.948 –0.526
49 0.623 0.736 –0.113 0.736 –0.113
50 0.747 0.838 –0.091 0.838 –0.091
51 0.976 0.986 –0.01 0.986 –0.010
52 1.508 0.93 0.578 0.930 0.578
53 0.544 0.228 0.316 0.228 0.316
54 0.000 –0.096 0.096 –0.096 0.096
55 0.477 0.805 –0.328 0.805 –0.328
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
(Property-Evaluation by Class Variables)
Copyright © 2011 SciRes. PP
279
56 2.301 1.944 0.357 1.944 0.357
57 1.362 2.045 –0.683 2.045 –0.683
58 0.000 –0.638 0.638 –0.638 0.638
59 Outlier
60 0.903 0.401 0.502 0.401 0.502
61 –0.155 –0.79 0.635 –0.790 0.635
62 0.602 0.062 0.54 0.062 0.540
63 1.969 1.588 0.381 1.587 0.382
64 Outlier
65 0.000 –0.039 0.039 –0.039 0.039
66 0.400 0.162 0.238 0.162 0.238
67 Outlier
68 –0.220 –0.865 0.645 –0.865 0.645
69 –0.680 –1.133 0.453 –1.133 0.453
70 –1.600 –1.395 –0.205 –1.395 –0.205
71 0.380 0.688 –0.308 0.688 –0.308
72 –0.072 –0.173 0.101 –0.173 0.101
73 0.550 0.064 0.486 0.064 0.486
74 –0.939 –1.003 0.064 –1.003 0.064
75 –0.701 –0.313 –0.388 –0.313 –0.388
76 –1.770 –1.131 –0.639 –1.131 –0.639
77 –0.903 –0.776 –0.127 –0.776 –0.127
78 –0.812 –1.17 0.358 –1.170 0.358
Table 6. List of virtual fragments.
S. No Fragment Specimen
Molecule Correlation F
1 C8H4O 1 –0.4294 174.1
2 NO2 19 –0.4006 147.1
3 Br 2 –0.3603 114.9
4 CO 1 0.3422 102.1
5 C9H4O4 56 0.3405 100.9
6 C6H4 5 –0.3177 86.4
7 C6H5O2 56 0.2928 72.2
8 C6H4O 1 0.2592 55.5
9 C8H5O4 63 0.2559 54.0
Figure 1. Correlation between observed and estimated log Ki
using 9-parametric model both from PRECLAV and NCSS
softwares.
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
280
(Property-Evaluation by Class Variables)
Figure 2. Residual plot for log Ki using 9-parametric model
both from PRECLAV and NCSS softwar e s.
Figure 3. Ridge plot.
Figure 4. Ridge plot.
5. Conclusions
From the results and discussion made above we conclude
that the PRECLAV software generates and proposes the
overall best model and that there is no need of perform-
ing successive or stepwise regression to arrive at the best
model. Such regressions are needed in NCSS software
for obtaining the best model. Furthermore, while using
PRECLAV software there is no need to perform model
validation separately. Finally, PRECLAV software pro-
poses virtual fragment which increases or decreases the
biological activity. From the comparison made above we
conclude that the PRECLAV software is the best for fu-
ture 2D-QSAR study.
6. Acknowledgements
One of the authors (Shalini Singh) is thankful to DST
(Department of Science and Technology) New Delhi for
the project grant and to the authorities of Bareily College,
Bareily-243001, UP, India for providing the facilities to
carryout this work. And (Bashirulla Shaik) is thankful to
Head, Department of Applied Sciences, NITTTR, Bhopal
for providing the Research facilities.
REFERENCES
[1] M. Karelson, “Molecular Descriptor in QSAR/QSPR,”
John Wiley & Sons, New York, 2000.
[2] PRECLAV software is purchased from Center of Organic
Chemistry (CCO)-Bucharest, Romanian Academy, Petru
Filip, pfilip@cco.ro.
[3] A. R. Katritzky, R. Petrukhin, S. Perumal, M. Karelson, I.
Prakash and N. Desai, “A QSPR Study of Sweetness Po-
tency Using the CODESSA Program,” Croatica Chemica
Acta, Vol. 75, No. 2, 2002, pp. 475-502.
[4] J. Devillers and A. T. Balaban, “Topological Indices and
Related Descriptors in QSAR and QSPR,” Gordon &
Breach Science Publishers, Amsterdam, 1999.
[5] M. Marder, H. Viola, J. A. Bacigaluppo, M. I. Colombo,
C. Wasowski, C. Wolfman, J. H. Medina, E. Ruveda and
A. C. Paladini, “Detection of Benzodiazepine Receptor
Ligands in Small Libraries of Flavone Derivatives Syn-
thesized by Solution Phase Combinatorial Chemistry,”
Biochemical and Biophysical Research Communications,
Vol. 249, No. 2, 1998, pp. 481-485.
doi:10.1006/bbrc.1998.9146
[6] K. Dekermendjian, P. Kahnberg, M. R. Witt, O. Sterner,
M. Nielsen and T. Liljefors, “The Use of a Pharma-
cophore Model For Identification of Novel Ligands for
The Benzodiazepine Binding Site of the GABAA Recep-
tor,” Journal of Medicinal Chemistry, Vol. 42, No. 21,
1999, pp. 4343-4350. doi:10.1021/jm991010h
[7] X. Hong and A. J. Hopffinger, “3D-Pharmacophores of
Flavonoid Binding at the Benzodiazepine GABAA Re-
ceptor Site Using 4D-QSAR Analysis,” Journal of Che-
Copyright © 2011 SciRes. PP
Modeling of the Interaction of Flavanoids with GABA (A) Receptor Using PRECLAV
(Property-Evaluation by Class Variables)
Copyright © 2011 SciRes. PP
281
mical Information and Modeling, Vol. 43, No. 1, 2003, pp.
324-326. doi:10.1021/ci0200321
[8] X. Huang, T. Liu, J. Gu, X. Luo, R. Ji, Y. Cao, H. Xue, J. T.
Wong, B. L.Wong, G. Pei, H. Jiang and K Chen, “3D-
QSAR Model of Flavonoids Binding at Benzodiazepine
Site in GABAA Receptors,” Journal of Medicinal Chemis-
try, Vol. 44, No. 12, 2001, pp. 1883-1891.
doi:10.1021/jm000557p
[9] P. Kahnberg, E. Lager, C. Rosenberg, J. Schougaard, L.
Camet, O. Sterner, E. Ostergaard Nielsen, M. Nielsen and
T. Liljefors, “Refinement and Evaluation of a Pharma-
cophore Model for Flavone Derivatives Binding to the
Benzodiazepine Site of the GABAA Receptor,” Journal of
Medicinal Chemistry, Vol. 45, No. 19, 2002, pp. 4188-
4201. doi:10.1021/jm020839k
[10] A. Da Settimo, G. Primofiore, F. Da Settimo, A. M. Marini,
E. Novellino, G. Greco, C. Martini, G. Giannaccini and A.
Lucacchini, “Synthesis, Structure-Activity Relationships,
and Molecular Modeling Studies of N-(Indol-3-ylglyoxy-
lyl)benzylamine Derivatives Acting at the Benzodiazepine
Receptor,” Journal of Medicinal Chemistry, Vol. 39, No.
26, 1996, pp. 5083-5091. doi:10.1021/jm960240i
[11] T. Blair and G. A. Webb, “Electronic Factors in the Struc-
ture-Activity Relations of Some 1,4-Benzodiazepin-2-
ones,” Journal of Medicinal Chemistry, Vol. 20, No. 9,
1977, pp. 1206-1210. doi:10.1021/jm00219a019
[12] G. Greco, E. Novellino, C. Silipo and A. Vittoria, “Study
of Benzodiazepines Receptor Sites Using a Combined
QSAR-CoMFA Approach,” Quantitative Structure-Acti-
vity Relationships, Vol. 11, No. 4, 1992, pp. 461-477.
doi:10.1002/qsar.2660110403
[13] S. P. Gupta and A. Paleti, “Quantitative Structure-Acti-
vity Relationship Studies on Benzodiazepine Receptor
Binding of Some Nonbenzodiazepine Series of Ligands,”
Quantitative Structure-Activity Relationships, Vol. 15, No.
1, 1996, pp. 12-16. doi:10.1002/qsar.19960150104
[14] P. R. Duchowicz, Ma. G. Vitale, E. A. Castro, J. C.
Autino, G. P. Romanelli and D. O. Bennardi, “QSAR
Modeling of the Interaction of Flavonoids with GABA (A)
Receptor,” European Journal of Medicinal Chemistry,
Vol. 43, No. 8, 2008, pp. 1593-1602.
doi:10.1016/j.ejmech.2007.11.009
[15] DRAGON Software for calculation of topological indices.
www.disat.unimib.it
[16] H. Kubinyi, “Variable Selection in QSAR Studies. II. A
Highly Efficient Combination of Systematic Search and
Evolution,” Quantitative Structure Activity Relationships,
Vol. 13, No. 4, 1994, pp. 393-401.
doi:10.1002/qsar.19940130403
[17] R. D. Cramer, D. E. Patterson and J. D. Bunce, “Compara-
tive Molecular Field Analysis (COMFA). 1. Effect of Sha-
pe on Binding of Steroids to Carrier Proteins,” Journal of
American Chemical Society, Vol. 110, No. 18, 1988, pp.
5959-5967. doi:10.1021/ja00226a005
[18] D. Chappel, “Trouble with CORBA,” 1998.
www.davidchappel.com.
http://www.davidchappell.com/articles/article_Trouble_C
ORBA.html
[19] OASIS. http://www.oasissolutionsllc.com/custom.html
[20] A. R. Katritzky, M. Karelson and R. Petrukhin, “Com-
prehensive Descriptors for Structural and Statistical Ana-
lysis”. http://www.codessa-pro.com/index.html
[21] A. Kovatcheva, G. Buchbauer and A. Golbraikh, “QSAR
Modeling of Alpha-Campholenic Derivatives with San-
dalwood Odour,” Journal of Chemical Information and
Computational Science, Vol. 43, No. 1, 2003, pp. 259-266.
doi:10.1021/ci020296n
[22] Hyperchem Molecular Modeling Software.
www.hyper. com
[23] M. J. S. Dewar and H. S. Rzepa, “Ground States of
Molecules. MNDO Results for Molecules Containing Be-
ryllium,” Journal of American Chemical Society, Vol.
100, No. 3, 1978, pp. 777-784. doi:10.1021/ja00471a020
[24] NCSS Statistical Analysis Software. www.ncss.com