African Journal of Biology and Medical Research

ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

DOCKING AND QSAR STUDIES OF NEW IMIDAZO [1,2-A] QUINOXALINE

DERIVATIVES USING GENETIC FUNCTION APPROXIMATION (GFA)
AGAINST HUMAN MELANOMA
Emmanuel Israel Edache1* and Shafiu Saidu2
1
Department of Pure and Applied Chemistry, University of Maiduguri, Brono State, Nigeria
2
Department of Chemistry, Kaduna State University, Kaduna-Nigeria
*
Email: [email protected], Tel: +2348066776802

ABSTRACT: In this paper, to comprehend the chemical-biological interactions governing

their activities toward antitumor activity, QSAR models of 31 derivatives of New imidazo [1,
2-a] quinoxaline derivatives with inhibitory tumor were developed. The quantitative structure-
activity relationship (QSAR) model was built by using the genetic function algorithm (GFA)
technique, and the best GFA model has SEE = 0.51748, R2 = 0.73038 cross-validated, R2adjusted
= 0.63234, F = 7.44967 (DF: 4, 11), and Q2 = 0.51664 non-cross-validated. The predictive
ability of the GFA model was further validated by a test set of 8 compounds, giving R2pred =
0.73038. Docking studies were used to discover the real conformations of chemicals in the
active site, as well as the binding mode shape to the binding site in enzyme. Ligand with
PubChem_CID number 44561182 has the least binding affinity with the enzyme. The
information provided by the 2D-QSAR model and docking may lead to a better understanding
of the structural requirements of 31 New imidazo [1, 2-a] quinoxaline derivatives and help to
design potential anti-tumor molecules.
KEYWORDS: Tumor, Quinoxalinamine Derivatives, GFA-MLR, QSAR, Applicability
Domain, Molecular Docking

INTRODUCTION
Regular cells in our bodies follow an organized path of development, separation, and death [1].
The tumor is a class of diseases categorized by out-of-control cell development or irregular
development of the cells with the ability to spread in the other parts of the body which damages
the body by forming bumps or multitudes of tissue called tumor [1, 2]. Despite increased
attention to tumor eradication, the disease causes more than a million deaths each year and is
the second cause of death in the world [3, 4]. In spite of several efforts in the handling of a
tumor, and the limitations that medications have, this disease became a big problem for the
health of societies [2]. Extensive laboratory suggestions from chemical, cell culture, and animal
studies indicate that antioxidants may slow or probably prevent the growth of tumors [5].
Quinazoline ring is a versatile lead molecule that has been studied widely which possesses
analgesic, anti-inflammatory, antitumor, antimicrobial, anticonvulsant, enzyme inhibition
activity, and many other activities [6, 7]. Several studies have shown that the Quinazoline
nucleus has potent activity against human cancer particularly by killing the cells in a tumor-
specific manner [8]. The nucleus has also been reported to have potent antioxidant activity [5].
Amino acids will reduce the side effects of the metabolite of the parent compound upon

67
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

metabolism in the body and enhance the solubility of the synthesized candidates when it is
incorporated into pharmacologically active quinazolinone moiety [9]. These observations gave
us a great motivation to the search for potential biological active drugs carrying new
imidazo[1,2-a] quinoxaline analogs which have been synthesized in good yields via a
bimolecular condensation of 2-imidazole carboxylic acid, followed by a coupling and
subsequent substitution on the imidazole ring by Suzuki Cross-coupling reaction using
microwave assistance [10]. Quantitative Structure-Activity Relationship (QSAR) modeling
and docking has become an extensively used tool [11] in computer-aided drug design (CADD).
CADD has been used to predictive environmental risk assessment and fate modeling, toxicity
and property prediction of chemicals and pharmaceuticals [12, 13] as well as in different
modeling problems in material sciences, analytical chemistry and
pharmacokinetics/pharmacodynamics profiling of new drug molecules [14]. Computer-aided
drug design (CADD), is utilized to achieve the desired results. CADD provides appreciated
perceptions into experimental findings and mechanism of action, new suggestions for
molecular structures to synthesize, and can help make cost-effective decisions before expensive
synthesis is started. Quantitative Structure-Activity Relationship (QSAR) modeling is a ligand-
based drug design method for both exploring and exploiting the relationship between chemical
structure and its biological action [15, 16, and 17]. To predict the activities of anticancer
compounds, quantum chemical descriptors like molecular orbital, dipole moment, charge, etc.
and molecular property descriptors like hydrophobic, steric coefficient, etc. have been applied
to develop 2D QSAR models [18, 19], while docking study is also performed to explore the
binding pocket in the enzyme and to understand the binding mode pattern for each compounds
[20].

MATERIALS AND METHODS

In the present paper, structure-cytotoxic activity relationships of a series of 31 compounds of
New imidazo[1,2-a] quinoxaline derivatives against human A375 cells (MTT) reported by
PubChem [21, 22] have been used for the 2D-QSAR model development. The Marvin Sketch
and Marvin View software package were used to building the chemical structures. Density
functional theory was adopted at the level of B3LYP/6-31+G (d, p) [23] to optimize the isolated
compounds using Spartan'14 version 1.1.4 software. Descriptor generation was performed
using the PaDel package version 2.20 [24]. DTC Lab (new_Dataset Division GUI v1.2
Division) software were used to divide the compounds into training and test set. The Kennard-
Stone algorithm was used to split data sets into two distinct subsets with an equal distribution
such that no sample from one subset should be too far from any sample of the other subset, and
the coverage should start on the boundary of the factor space [25, 26, 27, and 28], followed by
the application of statistical methods (GFA-MLR) to determine the main descriptors
responsible for the cytotoxic activity of the compounds under investigation. Nowadays, the
genetic function approximation (GFA) method [29] is considered superior to other variable
selection methods. It is a powerful optimization method that was inspired by evolutionary
principles, including survival of the fittest, reproduction, crossover, and mutation. In this study,
GA-MLR was used to build the QSAR model. The fitness function utilized herein was the
leave-one-out (LOO) cross-validated correlation coefficient (Q2). Measurement of cytotoxic
activity is expressed as half-maximal (50%) inhibitory concentration of a substance (IC50) and
values are expressed in Nano molar (nM – 10-9) levels. The values were converted to the pIC50
scale (-log IC50) to predict the narrow value wherein higher values indicate exponentially

68
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

greater potency. The pIC50 values were used as the dependent variables to construct the QSAR
model. The structures of these compounds and their PubChem_CID numbers are shown in
Table 1.
Table 1: Structures of dataset used for GA-MLR QSAR analysis with compound CID,
IUPAC CAS name and experiment IC50.

S/ Structure PubChem_CI IUPAC_CAS_NAME IC50

N D
1b 57469 1-(2-methylpropyl)-4- 70.
imidazo[4,5- 3
c]quinolinamine

2a 25253254 N-methyl-1-phenyl-4- 74.

imidazo[1,2- 1
a]quinoxalinamine

3c 25253255 1-(2-methoxyphenyl)-N- 0.6

methyl-4-imidazo[1,2- 5
a]quinoxalinamine

4b 25253256 1-(3-methoxyphenyl)-N- 3.4

methyl-4-imidazo[1,2- 7
a]quinoxalinamine

5a 25253257 1-(4-methoxyphenyl)-N- 0.3

methyl-4-imidazo[1,2-
a]quinoxalinamine

6b 44224743 1-(3-ethoxyphenyl)-N- 0.5

methyl-4-imidazo[1,2- 6
a]quinoxalinamine

69
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

7a 44224744 3-[4-(methylamino)-1- 1.2

imidazo[1,2- 8
a]quinoxalinyl]phenol

8a 44224745 1-(3-bromophenyl)-N- 173

methyl-4-imidazo[1,2-
a]quinoxalinamine

9a 44224966 N-methyl-1-[3- 122

(trifluoromethyl)phenyl]-
4-imidazo[1,2-
a]quinoxalinamine

10a 44224967 1-(3-chlorophenyl)-N- 0.2

methyl-4-imidazo[1,2-
a]quinoxalinamine

11b 44224968 3-[4-(methylamino)-1- 0.3

imidazo[1,2- 7
a]quinoxalinyl]benzoic
acid

12b 44224969 1-(3-fluorophenyl)-N- 2.1

methyl-4-imidazo[1,2- 9
a]quinoxalinamine

13a 44224970 3-[4-(methylamino)-1- 24.

imidazo[1,2- 9
a]quinoxalinyl]benzonitril
e

70
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

14a 44224971 N-methyl-1-(3- 27

nitrophenyl)-4-
imidazo[1,2-
a]quinoxalinamine

15b 44224972 1-(3-furanyl)-N-methyl-4- 40

imidazo[1,2-
a]quinoxalinamine

16c 104799 1-(2-chloroethyl)-3-(1- 31.

diethoxyphosphorylethyl) 6
-1-nitrosourea

17a 10244054 N-methyl-1-(2- 1.5

methylpropyl)-4- 7
imidazo[1,2-
a]quinoxalinamine

18a 24779760 N-methyl-1-(2- 2.3

phenylethyl)-4- 5
imidazo[1,2-
a]quinoxalinamine

19a 25254189 N,N-dimethyl-1-(2- 24

phenylethyl)-4-
imidazo[1,2-
a]quinoxalinamine

20a 25254190 1-(2-phenylethyl)-4- 66.

imidazo[1,2- 3
a]quinoxalinamine

71
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

21b 44561144 4-chloro-1-(2- 80.

phenylethyl)imidazo[1,2- 1
a]quinoxaline

22a 44561145 N,N-dimethyl-1-(2- 47.

methylpropyl)-4- 8
imidazo[1,2-
a]quinoxalinamine

23b 44561146 4-methoxy-1-(2- 400

phenylethyl)imidazo[1,2-
a]quinoxaline

24c 44561181 329

N-(2-chloro-6-
methylphenyl)-1-(2-
methylpropyl)-4-
imidazo[1,2-
a]quinoxalinamine

25c 44561182 N-(2-chloro-6- 85.

methylphenyl)-1-(2- 3
phenylethyl)-4-
imidazo[1,2-
a]quinoxalinamine

26a 44561183 N-methyl-4-imidazo[1,5- 121

a]quinoxalinamine

72
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

27c 44561184 N,N-dimethyl-4- 400

imidazo[1,5-
a]quinoxalinamine

28c 44561185 N-methyl-1-phenyl-4- 100

imidazo[1,5-
a]quinoxalinamine

29a 44561186 N,1-dimethyl-4- 78.

imidazo[1,5- 6
a]quinoxalinamine

30a 44561225 N-methyl-2-(2- 101

methylpropyl)-4-
pyrazolo[1,5-
a]quinoxalinamine

31c 44561226 N,N-dimethyl-2-(2- 173

methylpropyl)-4-
pyrazolo[1,5-
a]quinoxalinamine

Training seta; Test setb; outliersc

The relationship between dependent variable and independent variables was established by
GFA-multiple regression analysis using Material Studio. Significant descriptors were chosen
based on the statistical data analysis. Statistical quality of the generated QSAR equation was
judged based on the parameters like Friedman LOF and Cross-validation R-squared. Cross-
validation is a popular method used to test the reliability of QSAR models. In this study, leave-
one-out (LOO) method was applied to generate a number of improved data sets by removing
the first row and its value predicted using the rest of the data. Also, each row is left in turn, so
that the value of each row is predicted from all others. The model is judged based on these
predictions. Predictive ability of the generated model was assessed externally by predicting the
activities of test set. This condition may not be sufficient for a QSAR model to be truly
predictive [30]. An additional condition for high predictive ability of QSAR model is based on
external set cross-validation R2, (R2pred) and the regression of observed activities against
predicted activities and vice versa for validation set, if the conditions of Golbraikh and Tropsha

73
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

are satisfied [30]. Calculations relating to R2pred, R02 and the slopes, k and k' are based on
regression of observed values against predicted values and vice versa.
Molecular Descriptors
In spite of great advances in the field of drug design, the use of descriptors to define the
molecular structure of biologically active compounds is the key method used to discover new
lead molecules. Descriptors are the biochemical representative of a molecule in numerical
form, used for QSAR studies. The information encoded by descriptors normally depends on
the kind of molecular representation and the defined algorithm for its calculation. Some of
these include: topological indices, and geometrical, constitutional and physicochemical
descriptors. Constitutional descriptors are simple, commonly used descriptors reflecting the
molecular composition of a compound without any information about its topology. The most
common constitution descriptors are number of atoms, bond count, atom type, ring count, and
molecular weight (MW). These descriptors are inert to any conformation change and, therefore,
do not differentiate among isomers [31].
Domain of Applicability
Applicability domain of a QSAR model must be clear if the model is to be used for screening
new compounds. Predictive ability of the model may be considered reliable if the data set falls
into this domain [32, 33, and 34]. One simple approach is based on Euclidean based and the
calculation of leverages for compounds are used in the study.
Y-randomization
Another simple way of proving that the structure-activity relationships established in this study
do not result from coincidence involves checking their robustness by means of the so-called Y-
randomization [35]. This test ensures the robustness of a QSAR model [32] and to assess the
multiple linear regression models obtained by descriptor selection. In y-randomization test, the
dependent variable (pIC50) is randomly shuffled and a new QSAR model is developed keeping
X-data (descriptors) intact. The new models are expected to have low R2 and Q2 values, which
determine the statistical significance of the original model.
Molecular Docking
Docking studies were performed to further explain the results of our 2D-QSAR study and
explore the possible binding modes of these compounds. Docking is often utilized to forecast
the binding orientation of small molecules (drug candidates) to their macromolecular target
(such as protein, carbohydrate and nucleic acid) with the aim to determine their tentative
binding parameters [36]. This establishes raw data for the rational drug designing (structure-
based-drug development) of new agents with better efficacy and more specificity [37]. The
research in the drug discovery process involves virtual screening (VS) which is a computational
technique used for the rapid exploration of huge collections of chemical structures in order to
recognize those structures that are most likely to bind to a drug target, usually a protein receptor
or enzyme [38]. Identification of possible protein targets of small chemical molecules is a main
step for unravelling their primary causes of actions at the molecular level [39]. Docking studies
were employed to locate the appropriate binding orientations and conformations of these New
imidazo[1,2-a] quinoxaline derivatives interacting with ligand using the docking program
Molegro Virtual Docker, iGemdock and PyRx. By default, the docking program produces

74
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

docked structures for each New imidazo[1,2-a] quinoxaline derivatives. The conformation with
the lowest docking energy in the most inhabited cluster is selected as the possible 'active'
conformation against the protein active site. The X-ray crystal structure, taken from the Protein
Data Bank (pdb: 6EBE) was used to dock. At the beginning of docking, all the water molecules
were removed and hydrogen atoms added to the protein were applied. The structures were
docked by the active site defined through a grid box (Vina Search Space: center_x = -9.7952,
center_y = -1.7878, center_z = 16.0475), Dimensions (Angstrom): size_x = 47.5025488281,
size_y = 44.5045238495, size_z = 54.4464433479 and exhaustiveness = 8.

RESULTS AND DISCUSSIONS

pIC50 = -7.20917 - 1.4154ALogP - 0.06171AMR +2.92517SC-3 + 3.74747maxaasC ----Model 1
Description about selected variables are as follows:
ALogP (PaDEL; 2D): Ghose-Crippen LogKow
AMR (Dragon; Molecular properties): Ghose-Crippen molar refractivity
SC-3(PaDEL; 2D): Simple cluster, order 3
MaxaasC (PaDEL; 2D): Maximum atom-type E-State: C: -
SEE :0.51748, r^2 :0.73038, r^2 adjusted :0.63234, F :7.44967 (DF :4, 11), Q2 :0.51664, LOF:
1.2666, PRESS :5.2809, SDEP :0.5745, r^2 :0.73038, r0^2 :0.73038, reverse r0^2:0.67893,
rm^2(test) :0.73038, reverse rm^2(test) :0.56471, average rm^2(test) :0.64755, delta rm^2(test)
:0.16567, rmsep:0.42907, rpred^2 :0.73038, Q2f1 :0.73038, Q2f2 :0.73038
Some External Validation Parameters
Without scaling After scaling
rm^2(overall): 0.59546 0.57463
reverse rm^2(overall): 0.47544 0.48104
average rm^2(overall): 0.53545 0.52783
delta rm^2(overall): 0.12001 0.09359
Golbraikh and Tropsha acceptable model criteria’s:
1. Q^2 0.51664, Passed (Threshold value Q^2>0.5)
2. r^2 0.73038, Passed (Threshold value r^2>0.6)
3. |r0^2-r'0^2| 0.05145, Passed (Threshold value |r0^2-r'0^2|<0.3)
4. k = 1, [(r^2-r0^2)/r^2] = 0 OR k' = 0.86976, [(r^2-r'0^2)/r^2] = 0.07044 Passed

75
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

(Threshold value: [0.85<k<1.15 and ((r^2-r0^2)/r^2) <0.1] OR [0.85<k'<1.15 and ((r^2-

r'0^2)/r^2) <0.1])
Table 2: GA-MLR QSAR analysis with corresponding Experimental and predicted class
of tumour inhibitors.

PubChem_CID
(Train) Yobs Ypred (Residual)^2 (Yobs-Ybar)^2
44224745 0.187087 -0.36463 0.304392 1.085594
44224966 -0.10721 -0.39265 0.081477 0.558938
44224967 -0.25042 -0.04276 0.043121 0.365314
44224970 -0.54033 -0.27352 0.071186 0.098912
44224971 -1.3962 -1.58743 0.036567 0.293079
10244054 -1.43136 -0.99634 0.189246 0.332389
25254189 -1.86982 -0.73271 1.293012 1.030197
25254190 -1.49969 -1.38916 0.012215 0.415838
44561145 -0.37107 0.182161 0.306062 0.234028
44561183 -1.38021 -1.45988 0.006347 0.276024
24779760 -1.67943 -1.261 0.175082 0.679959
44561186 -1.93095 -2.97387 1.087687 1.158028
44561225 -0.1959 -0.64536 0.202017 0.434192
25253254 -1.89542 -1.54185 0.125011 1.082829
25253257 0.431798 -0.66553 1.204123 1.655417
44224744 0.251812 -0.1268 0.143351 1.22466
PubChem_CID
(Test) Yobs Ypred (Residual)^2 (Yobs-Ybar)^2
57469 -1.84696 -3.39226 2.387952 0.984309
44224968 -1.60206 -1.56728 0.00121 0.55835
44224969 -1.90363 -0.17617 2.984113 1.099983
44224972 -1.82151 -1.005 0.6667 0.934474
44561144 -2 -1.75506 0.059994 1.31141
44561146 -0.34044 -0.50363 0.026629 0.264595
25253256 0.69897 -0.62523 1.753496 2.4143
44224743 0.522879 -1.20436 2.983337 1.898086

A total of 24 compounds (Table 1) were used for the QSAR model generation. It is essential to
assess the predictive power of models by using a test set of compounds. This was achieved by
setting aside 8 compounds as a test set such that it represented the various functional groups
included in the training set and had a regularly distributed biological data. The mean of the
biological activity of the training and test set was -0.9049 and -0.9364, respectively. The quality
factor (Q) was performed to access the robustness and statistical confidence. The higher value
of R2, RMSEP, Q, and F and lower value of Se, and RMSECV of Model 1 revealed that Model
1 was robust and promising. In the developed Model the value of the coefficient of correlation
was significantly high supporting reliability and goodness. The accuracy of the Model 1 was
ascertained by correlation coefficient (R2 = 0.7304), statistical significance more than 95%
(against tabulated value F = 3.4035) and low standard error of estimate (0.5175). The model
shows that parameter SC-3 and maxaaC showed a positive contribution. The regression model

76
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

has small residuals that can be seen in (Table 2). LOO cross-validation analysis revealed that
R2-Q2LOO < 0.3 (0.7304 – 0.5166 = 0.21). The robustness of the model was justified According
to Golbraikh and Tropsha [30], the proposed QSAR model is predictive as it satisfies this
conditions like R2pred > 0.5, R2 > 0.6, r2-r2o/r2 < 0.1, r2-r'2o/r2 < 0.1 and 0.85 ≤ K ≤ 1.15 or 0.85
≤ K′ ≤ 1.15 , but this model satisfy the following criteria R2pred = 0.7304 > 0.5, and R2 = 0.7304
> 0.6. So, this QSAR model is predictive as it's satisfied this condition reported by Golbraikh
and Tropsha, [30]. The model also possesses a high value of FIT criterion [40, 41], the internal
validation parameter of the model (Q2cv = 0.5166) was also good. It can be observed that the
obtained model has sensible internal and external value. Nevertheless, it is always necessary to
obtain a model that can relate the physicochemical properties represented by the selected
molecular descriptors to the action mechanism of the system under study [42]. The estimation
of probable error of the coefficient of correlation (PE) is another requirement for validating the
method [43]. This is defined as PE = 2/3(1-R2/√n): Where R in multiple correlations is the
correlation coefficient and n is the number of compounds under study. It is argued that: (1) if
R < PE, then R is not significant; (2) if R > PE, several times; at least 3-times greater correlation
is indicated, and (3) if R > 6PE, then the correlation is good. The 6PE = 0.269617 indicates
that the proposed correlations are good.

2
1.5
PubChem_CID
1 57469
Standard Residuals

0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
-0.5 1.252236123 -
0.653781848
-1
-1.5
-2
-2.5
Leverages

training set test set

Fig. 1: The Williams plot, the plot of the standardized residuals vs. leverages

Figure 1 shows the standardized residuals (σ) versus the leverage samples plot, and it was used
for the identification of outliers. No compound presented residuals higher than 2.5σ. Only one
compound presented leverage higher than the leverage cutoff line, but it can be considered
acceptable [44]. Therefore, the model can be considered free of outliers, something which
guarantees the maximum possible representation in terms of structure and range of inhibitory
activity for the dataset under study. Euclidean based applicability domain helps to specify the
scope of their proposed models, therefore, defining the model limitations concerning its
structural domain and response space. If an external compound is beyond the defined scope of
a given model, it is considered outside that model's Applicability Domain (AD) and cannot be
associated with a reliable prediction. The resulting model can be reliably applicable for only

77
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

those compounds which are inside this domain. Euclidean based applicability domain helps to
ensure that the compounds of the test set are representative of the training set compounds used
in model development. It is based on distance scores calculated by the Euclidean distance
norms. At first, normalized mean distance scores for training set compounds are calculated and
these values range from 0 to 1(0 = least diverse, 1 = most diverse training set compound). Then
normalized mean distance score for test set are calculated (Table 3), and those test compounds
with scores outside 0 to 1 range are said to be outside the applicability domain. This can also
be checked by plotting a 'Scatter plot' (normalized mean distance vs. respective activity)
including both training and test set as shown in Figure 2. If the test set compounds are inside
the domain/area covered by training set compounds that means these compounds are inside the
applicability domain otherwise not [32, 34].
Table 3: Euclidean based applicability domain (AD) for Model 1

PubChem_CID Normalized Mean

(Training Set) Distance Score Mean Distance Distance
44224745 70.48381 4.405238 0.058052
44224966 68.54478 4.284049 0.043204
44224967 64.68526 4.042829 0.01365
44224970 64.30307 4.018942 0.010723
44224971 102.1794 6.386216 0.30076
10244054 62.90267 3.931417 0
25254189 93.37212 5.835757 0.233318
25254190 193.4945 12.09341 1
44561145 75.42328 4.713955 0.095876
44561183 64.86008 4.053755 0.014989
24779760 67.26472 4.204045 0.033402
44561186 114.4197 7.151229 0.394489
44561225 76.78591 4.79912 0.10631
25253254 193.4887 12.09304 0.999955
25253257 63.43943 3.964964 0.00411
44224744 84.76635 5.297897 0.16742
PubChem_CID Distance Score Mean Distance Normalized Mean
(Test Set) Distance
57469 123.8296 7.739348 0.466544
44224968 74.70109 4.668818 0.090346
44224969 144.0752 9.004703 0.621575
44224972 284.4916 17.78073 1.696805
44561144 64.35608 4.022255 0.011129
44561146 114.4033 7.150205 0.394363
25253256 63.47061 3.966913 0.004349
44224743 103.5443 6.471521 0.311211

78
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

PubChem_CID 4422492
Normalized mean distance
1.8
1.696805
1.6
1.4
1.2
1
0.8
0.621575 0.394363 0.6
0.466544 0.4
0.311211
0.011129 0.2
0.090346
0 0.004349
-2.5 -2 -1.5 -1 -0.5 0 0.5 1
Experimental pIC50
training test

Fig. 2: Euclidean Based applicability domain plot, the plot of the normalized mean
distance vs. observed IC50

In addition to Euclidean based applicability domain (AD), the internal predictive ability of the
model was further assessed by a Y-randomization performed with 16 derivatives for 10 times.
The average of 10 readings was given as average Q2 as shown in Table 4; the Y-randomization
test (Table 4) ensures the robustness of a QSAR model [32] and to assess the GFA models
obtained by descriptor selection [45]. In the Y-randomization test, the dependent variable is
randomly shuffled and a new QSAR model is developed keeping molecular descriptors intact.
The new models are expected to have low R2 and Q2LOO values, which determine the statistical
significance of the original model. The low R2 and Q2LOO values of the random models shown
in Table 4 and the value of R2p = 0.5894 (R2p≥0.5) indicates that there is no chance of
correlation or structural dependency in the proposed model. Consequently, model 1 can be
considered as a perfect model with both high statistically significant and excellent predictive
ability.
Table 4: The average R, R2 and Q2LOO values after several Y-Randomization

Model 1 R R^2 Q^2

Original 0.854625 0.730383 0.51664
Random 1 0.344168 0.118452 -1.21506
Random 2 0.299327 0.089597 -1.58454
Random 3 0.411688 0.169487 -1.23794
Random 4 0.608113 0.369801 -0.36964
Random 5 0.459707 0.211331 -0.76204
Random 6 0.730684 0.5339 -0.99338
Random 7 0.371123 0.137732 -1.61845
Random 8 0.62597 0.391838 -0.00437
Random 9 0.550345 0.30288 -0.33447
Random 10 0.646824 0.418382 -0.7894

79
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

Random Models Parameters Results

Average r : 0.504795
Average r^2 : 0.27434
Average Q^2 : -0.89093
cRp^2 : 0.58936

Interpretation of descriptors
By understanding the descriptors contained in the QSAR model, it is possible to gain certain
insights into issues, which are related to the antitumor activity. For this reason, an acceptable
interpretation of the selected descriptors is provided below. The brief descriptions of
descriptors are shown in model 1 and Table 5. To observe the significance as well as the
involvement of each descriptor in the model, the value of the mean effect (MF) was calculated
for each descriptor [46, 47]. The MF value shows the relative importance of a descriptor, related
with the other descriptors in the model. Its sign designates the variation direction in the values
of the activities as a result of the increase (or decrease) of the descriptor values.
Table 5: Variance Inflation Factor (VIF) and Mean effect (MF) of each Descriptors

Descriptors VIF MF
AlogP 2.030022475 -0.006133863
AMR 1.328239136 -0.23656501
SC-3 2.401363209 0.593332196
maxaasC 2.580941216 0.649366678

AlogP –thermodynamic descriptor. AlogP is the partition coefficient calculated using atom-
based approach and represents the hydrophobicity of the molecules [48]. A negative mean
effect of this descriptor illustrates that the activity increases with decreasing the value of AlogP,
which means that the partition coefficient calculated using atom-based approach and represents
the hydrophobicity of the molecules will benefit the activity.
Quantum mechanical calculations have become routine even for large molecular systems and
therefore the information related to the structure and electronic distribution can be easily and
efficiently used in deriving new descriptors and explaining the properties of molecules. AMR
Ghose-Crippen molar refractivity, a quantum mechanical descriptor, a measure of the total
polarizability of a mole of a substance, can be estimated, is a common molecular descriptor
accounting for molecular size and polarizability [49]. The negative sign of this descriptor
(Table 2) indicates that the pIC50 value is indirectly related to this descriptor. Hence, it was
concluded that by decreasing the molecular size and polarizability, the value of this descriptor
decreased, causing an increase in its pIC50 value.
The selected parameter SC-3 (PaDEL; 2D): Simple cluster, order 3 encode structure
information specifically based on a branch point, emphasizing the immediate branch point
environment. SC-3 exhibits the second largest positive mean effect (MF) Table 5 (0.5933) to
pIC50. Since the value of SC-3 (PaDEL; 2D): Simple cluster, order 3 is positive, increases in
values of SC-3 (PaDEL; 2D): Simple cluster, order 3 are advantageous to improving the
antitumor activity.

80
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

The desirability and suggested advantage of E-states over simple counts of the equivalent atom
types is that E-states values for each atom in a given molecule 'reflect' the steric and electronic
effects of the surrounding atoms and as such, could be best described as information-rich
atomic descriptors [50, 51]. Electro topological state atom type descriptor MaxaasC represents
Maximum atom-type E-State: C: -; this descriptor contributes positively MF (Table 5) which
indicates that inhibitory activity of New imidazo [1, 2-a] quinoxaline derivatives will increase
with Maximum atom-type E-State: C: -. The corresponding VIF values of the four descriptors
are presented in Table 5. As can be seen from this table, all the variables have VIF values of
less than five, indicating that the obtained model has statistical significance, and the descriptors
were found to be reasonably orthogonal [52, 53].
Docking Analysis
One application of molecular docking is to design therapeutic in-silico by optimizing targeted
lead candidates against the protein. The lead candidates can be found using a docking algorithm
that aims to identify the optimal binding mode of a small molecule (ligand) to the active site of
the macromolecular target. Twenty-four New imidazo [1, 2-a] quinoxaline derivatives to obtain
more effective compounds as inhibitors of the tumor. Molegro virtual docker (MVD) was used
to predict various orientations or conformations of the drugs against the protein targets. The
conformations with the least binding energies were selected and saved. The average binding
energies were calculated for each ligand after ten simulations with MVD. The Hydrogen bond
score, Number of Hydrogen bond, and interacting residues of the protein with the ligands were
also analyzed using the software. In the case of the crystal structures PDB code: 6EBE
complexes [54], the program generally identified three different binding sites (Figure 3).

Fig. 3: The three cavity MVD-detected cavities in anti-tumor, (PDB code 6EBE),
detected cavity green, carbon atoms grey, oxygen atoms red, nitrogen atoms blue.

Additionally, docking of these ligands, New imidazo [1, 2-a]quinoxaline derivatives was
performed with the crystal structure and each molecule selects the best position to define the

81
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

re-rank score. In each docking run, the best poses were selected based on their MVD re-rank
scores, and the mean of the six re-rank scores were then computed as the final score for each
molecule. The MVD score and the re-rank scores of the best poses for each of the docking
studies New imidazo[1,2-a] quinoxaline derivatives with protein PDB code: 6EBE are
summarized in Table 6.
Table 6: Plant Score, MolDock Score and Re-rank score (kcal/mol) for New imidazo[1, 2-
a]quinoxaline derivatives when docked with 6ebe crystal structure

Pubchem_CID Plant Score MolDock Score Re-rank Score

(Ligand)
44561182 -80.1715 -115.862 -88.8733
44561144 -70.2567 -82.5739 -61.3965
25254189 -70.2156 -84.4104 -65.7448
25254190 -69.892 -92.7031 -67.7022
25253257 -63.4977 -100.392 -77.4534
44561145 -61.8085 -78.6937 -65.414

Furthermore, the obtained score is between -78.6937 and -115.862cal/mol. Moldock score of
target compound PubChem_CID 44561182 is lower than PubChem_CID 44571145.
PubChem_CID 44561182 is a proven active compound as an inhibitor of anti-tumor protein
PDB code: 6EBE. By blocking PBD binding to its recognition motif, PubChem_CID 44561182
disrupted the human anti-tumor protein PDB code: 6EBE subcellular localization and
eventually arrested the cell cycle [55]. The docking results compound with 6EBE reveals no
electrostatic interactions but it has a hydrogen bonding and steric interaction between the ligand
to the receptor. The key residues in the binding pocket were Gly233, Glu236, Lys170, Asn232,
Phe231, Asn61, Leu60, Ile167, Lys172, Thr169, and Gly171. The ligand formed hydrogen
bonding (H-bond) with Gly171 in the binding site. This result indicates that the above-
mentioned molecules are predicted to be an antitumor drug candidate. The superposition of
ligand as observed in the cavity of the crystallographic structure of 6EBE and the best
conformation obtained theoretically are shown in Fig. 4. The result suggests that the software
reproduced the appropriate conformation of PubChem_CID 44561182 inside its binding site in
the protein (PDB code: 6EBE), the anion binding pocket is blue and specificity pocket is red
Fig.4c.

82
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

a. Hydrogen Bonds b. Steric interactions & interaction

Overlay

c. binding pocket

Fig.4. the best score docking solution of compound 44561182 with the selected crystal
structure of 6ebe. Amino acids in the active site are presented in lines and ligand is
presented in lines with fix color blue lines represent the hydrogen bonds between the
ligand and the active site of 6ebe.

The basic of iGEMDOCK is the GEMDOCK, which is a robust and well-developed tool. By
means of iGEMDOCK, the predicted poses produced from the GEMDOCK can be directly
visualized by a molecular visualization tool and analyzed by post-analysis tools. iGEMDOCK
offers the post-analysis tools by using k-means and hierarchical clustering methods based on
the docked poses (i.e. protein-ligand interactions) and compound properties (i.e. atomic
compositions). We validated the protein-ligand docking accuracy and screening accuracies of

83
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

iGEMDOCK by using a training and test set with 24 protein-ligand complexes. Based upon
the least binding affinity and other parameters, the best pose was selected and the docked
structures were visualized in Discovery Studio and Melogro molecular viewer software (Fig.
5a and 5b) for detailed residue-ligand interactions. The best inhibitor pose energy was obtained
-93.6494kcal/mol with a hydrogen bond of -7kcal/mol between the inhibitor and enzyme.
Docking of PubChem_CID 44561182 with protein represents that ligands were effectively
bound by interacting with Leu189, Leu47, Tyr191, Ser259, and Asp19 which are the residues
in the active site of protein (PDB code: 6EBE). The detailed docking results are shown in Table
7.
Table 7. iGEMDOCK Result: Computationally predicted potencies of 11 compounds
screened

Cpd CID (Ligand) Energy VDW HBond

25253256 -87.5408 -87.5408 0
25253257 -89.5498 -74.7659 -14.7839
44224743 -85.6956 -85.6956 0
44224970 -81.7628 -81.7628 0
10244054 -80.5062 -80.5062 0
25254189 -86.7377 -83.2393 -3.49839
25254190 -92.7138 -80.7486 -11.9652
44561144 -84.595 -78.595 -6
44561145 -84.6749 -73.3491 -11.3258
44561181 -89.203 -82.203 -7
44561182 -93.6494 -86.6494 -7

a. Visualization of Docking between 6ebe b. Visualization of Docking

and compound CID 44561182. between 6ebe and compound CID
44561182.

Fig. 5. Visualization of Docking between 6ebe and compound CID 44561182 using
Discovery Studio and Melogro molecular viewer respectively.

84
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

PyRx is an open-source package to accomplish virtual screening. It is a combination of several

software such as AutoDock Vina, AutoDock 4.2, Mayavi, Open Babel, etc. PyRx uses Vina
and AutoDock 4.2 as docking software.
Table 8. PyRx AutoDock Vina Result of MVD and iGEMDOCK selected compounds
(kcal/mol)

PuBChem_CID (Ligand) Result PuBChem_CID (Ligand) Result

25254190 -8.1 - -
44561182 -8.6 10244054 -6.7
44561144 -7.1 44561181 -7.8
25254189 -8.2 44224970 -9.0
25253257 -7.5 44224743 -7.7
44561145 -6.8 25253256 -7.9

The results of PyRx docking experiments of tumor inhibitors using AutoDock Vina are
summarized in Table 8. For each docking experiment, the lowest energy docked conformer
was selected from 10 runs. Ligand 44224970 (Fig. 6) showed better inhibition potential than
ligand 44561182 (Fig. 7), a potent tumor inhibitor, with binding energy -9.0 kcal/mole (Table
8). Modeling and docking analysis revealed the nature of the active site and some key
interactions that enabled the binding of inhibitor ligand 44561182 to the active site. The
numbers of the binding modes (compound CID 44561182) were thirteen (n=13) while
compound CID 44224970 was twelve (n = 12), respectively. Based on the present molecular
docking study, ligand 44561182 appeared as a strong binder to the enzyme (6ebe) active site
than the ligand 44224970 and the interacting numbers of amino acids and conventional
hydrogen bonds might be critical factors for regulating target protein activity. These data also
suggest that computer-aided drug design process using Melogro Virtual Docker, iGemDock
and PyRx tools is highly reliable and can be a good example for identifying the action
mechanism between the 6ebe (enzyme) and its interacting ligands.

Fig. 6: Binding interaction of compound CID 44561182 with antitumor (PDB ID-6EBE)

85
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

Fig. 7: Binding interaction of top ten least energetic molecules (Ligand 44224970) with
anti-tumor (PDB ID-6EBE)

CONCLUSION
Present work aimed to develop a 2D QSAR model for imidazo[1,2-a] quinoxaline to identify
the novel molecules that have well in silico predictions (QSAR and docking). GFA method
was used for variable selection and model building of imidazo[1,2-a] quinoxaline derivative.
The successful model was built with descriptors AlogP, AMR, SC-3, and maxaasC having
SEE: 0.51748, r^2:0.73038, r^2 adjusted: 0.63234, F: 7.44967 (DF: 4, 11), Q2:0.51664. The
cross-validation method, Y-randomization technique, applicability domain, and external
validation indicated that the model is statistically significant and has good internal and external
predictability. The reliability of these predictions for screened molecules, which were not part
of the QSAR training set, was also assessed by domain of applicability (Leverage and
Euclidean based applicability domain). Most of the molecules were followed the same domain
as a training set; hence, predictions were reliable. The 2D descriptors were related to topology
and 3D arrangement of atoms in molecules that can be used to design new inhibitors with good
potency.
The docking results of screened molecules ligand 44561182 were exhibited consistency in
terms of position into the active site and binding modes in all docking runs. As concluded from
these studies, imidazo[1,2-a] quinoxaline using known active enabled to identify new hits with
good in silico activities and binding poses, but in vitro assay to verify their experimental
activity need to be done.

86
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

Conflict of interest
The authors declare that there is no conflict of interest regarding the publication of this paper.
Also, they declare that this paper or part of it has not been published elsewhere.
Acknowledgments
The authors gratefully acknowledge the Department of Chemistry, Ahmadu Bello University,
Zaria (Samaru, Zaria-Nigeria); for computational studies.

REFERENCES
[1] Abechi, S. E and Edache, E. I. IJPSR. 2016, 7(04), 204-220.
[2] Agarwal, S.; Mehrotra, R. JSM Chem. 2016, 4(2), 1024.
[3] Ansari, K. I; Kasiri, S; Grant, J. D; Mandal, S. S. J. of Biomol. Screening, 2010, 16,
26–35.
[4] Chen, J.; Shen, Y.; Liao, S.; Chen, L.; Zheng, K. Int. J. Quantum Chem. 2007, 107,
1468–1478.
[5] Cherkasov, A.; Muratov, E. N.; Fourches, D.; Varnek, A.; Baskin, I. I.; Cronin, M.;
Dearden, J.; Gramatica, P.; Martin, Y. C.; Todeschini, R. J. Med. Chem. 2014, 57,
4977–5010.
[6] Clark, D. E. Expert Opin Drug Discov. 2006, 1(2), 103–110.
[7] Cooper, A.; Potter, T.; Luker, T. Drug Metab. Dispos. 2010, 38, 2218–2225.
[8] Danishuddin and Khan, A.U. Drug Discovery Today, 2016, 21.
[9] Deleuze-Masquefa, C.; Moarbess, G.; Khier, S.; David, N.; Gayraud-Paniagua, S.;
Bressolle, F.; Pinguet, F.; Bonnet, P. A. Eur. J. Med. Chem. 2009, 9, 3406-11. doi:
10.1016/j.ejmech.2009.02.007.
[10] De-Melo, E. B. Scientia Pharmaceutica. 2012, 80, 265–281
doi:10.3797/scipharm.1112-21.
[11] Dimitrov, S.; Dimitrova, G.; Pavlov, T.; Dimitrova, N.; Patlewicz, G.; Niemela, J.;
Mekenyan, O. A. J. Chem. Infor. Modeling. 2005, 45, 839–49.
[12] Golbraikh, A.; Tropsha, A. J. Comput. Aided Mol. Des. 2002, 16, 357–369.
[13] Gramatica, P. QSAR Comb Chem. 2007, 26, 694–701.
http://dx.doi.org/10.1002/qsar.200610151.
[14] Guedes, I. A.; de Magalhães, C. S.; Dardenne, L. E. Biophysical Reviews. 2014, 6,
75-87.
[15] Hall, L. H.; Mohney, B.; Kier, L. B. J Chem Inf Comput Sci. 1991, 31, 76-82.
[16] Hassan, G.; Dashtbozorgi, Z.; William E. A. Jr. Eur. J. Pharmaceutical Sci. 2012,
47(2), 421-429.
[17] Holland, J. H. Adaptation in Natural and Artificial Systems, University of Michigan
Press, Ann Arbor, 1975.
[18] Ibezim, E.; Duchowicz, P. R.; Ortiz, E.V.; Castro, E. A. Chemometrics and Intelligent
Laboratory Systems. 2012, 110, 81–88.
[19] Jaiswal, M.; Khadikar, P.V.; Scozzafava, A.; Supuran, C. T. Bioorg. Med. Chem.
Lett. 2004, 14, 3283–3290.
[20] Jaworska, J.; Nikolova-Jeliazkova, N.; Aldenberg, T. Alternative Laboratory Animals.
2005, 33, 445–459.

87
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

[21] Kar S, and Roy K, Predictive toxicology using QSAR: a perspective, J. Indian Chem.
Soc., 2010, 87, 1455–1515.
[22] Kennard, R.W.; Stone L. A. Technometrics, 1969, 11, 137– 148.
[23] Khan, N. U. H.; Pandya, N.; Kureshy, R. I. Mol. and Biomol. Spectroscopy. 2009,
74(1), 113–119.
[24] Kier, L. B.; Hall, L. H. Molecular Structure Descriptors: The Electrotopological State.
Academic Press: New York; 1999.
[25] Kubinyi, H. Quant Struct Act Relat. 1994, 13, 285-294.
[26] Kubinyi, H. Quant Struct Act Relat. 1994, 13, 393-401.
[27] Kumar, P.; Agarwal, J. C.; Nath, C.; Bhargava, K. P.; Shanker, K. Pharmazie, 1981,
36, 780.
[28] Kusumaningrum, S.; Budianto, E.; Kosela, S.; Sumaryono, W.; Juniarti, F. JAPS.
2014, 4 (11), 047-053, DOI: 10.7324/JAPS.2014.4119
[29] Liao, C.; Sitzman, M.; Pugliese, A.; Nicklaus, M. C. Future Med. Chem. 2011, 3,
1057–1085.
[30] Meyyanathan, M.; Chandrasekhar, K. Indian Drugs, 2006, 43(6).
[31] Moarbess, G.; Deleuze-Masquefa, C.; Bonnard, V.; Gayraud-Paniagua, S.; Vidal, J.
R.; Bressolle, F.; Pinguet, F.; Bonnet, P. A. Bioorg Med Chem. 2008, 16(13), 6601-
10. doi: 10.1016/j.bmc.2008.05.022.
[32] National Center for Biotechnology Information. PubChem Database.
Source=ChEMBL, AID=1315713, https://pubchem.ncbi.nlm.nih.gov/bioassay/
433907 (accessed on June 30, 2019).
[33] Nocentini, A.; Trallori, E.; Singh, S.; Lomelino, C. L.; Bartolucci, G.; Di Cesare
Mannelli, L.; Ghelardini, C.; McKenna, R.; Gratteri, P.; Supuran, C. T. J. Med. Chem.
2018, 61, 10860-10874.
[34] Noolvi, M. N.; Patel, H. M.; Bhardwaj, B. Digest J. Nanomater. Biostruct. 2010, 5,
387–401.
[35] Noolvi, M. N.; Patel, H. M.; Bhardwaj, V.; Chauhan, A. Europ. J. Med. Chem. 2011,
46, 2327e2346
[36] Pereira, F.; Latino D. A. R. S.; Gaudêncio, S. P. Molecules 2015, 20, 4848-4873;
doi:10.3390/molecules20034848.
[37] Pourbasheer, E.; Riahi, S.; Ganjali, M. R.; Norouzi, P. Eur. J. Med. Chem. 2009, 44,
5023–5028.
[38] Rajasekaran, S; Gopalkrishna, R. Der Pharmacia Lett. 2012, 4(1), 349-353.
[39] Rekha, S. H.; Basavaraj, R. P.; Dayananda, S. B.; Ramesh, S. V.; Kalagouda B. G.
Der Pharma Chemica. 2010; 2(2), 116-128.
[40] Riahi, S.; Pourbasheer, E.; Ganjali, M. R.; Norouzi, P. J. Hazard. Mater. 2009, 166,
853-859
[41] Roy, K. Expert Opin. Drug Discov. 2007, 2, 1567–1577.
[42] Roy, K.; Kar, S.; Ambure, P. Chemometrics and Intelligent Laboratory Systems,
2015, 145, 22–29.
[43] Shapiro, S.; Guggenheim, B. Quant. Struct.-Act. Relat. 1998, 17, 327–337.
[44] Singh, J.; Singh, S.; Thakur, S.; Lakhwani, M.; Khadikar, P. V.; Agrawal, V. K.;
Supuran, C.T. ARKIVOC 2006, (xiv), 103-118.
[45] Snarey, M.; Terrett, N. K.; Willet, P.; Wilton, D. J. J. Mol. Graphics Modell. 1997,
15, 372– 385.
[46] Somanathan, R. R.; Hellberg, L. H. Synth. Commum, 1998, 28(1), 2077-2086.
[47] Tropsha, A.; Gramatica, P.; Gombar, V. K. QSAR Comb Sci. 2003, 22, 69–77.

88
African Journal of Biology and Medical Research
ISSN: 2689-534X
Volume 3, Issue 3, 2020 (pp. 67-89) www.abjournals.org

[48] Vaidya, A.; Jain, A.K.; Kumar, B.R.P.; Sastry, G.N.; Kashaw, S.K.; Ram Kishore
Agrawal, R.K. Arabian J. Chem. 2017, 10, S3936–S3946.
http://dx.doi.org/10.1016/j.arabjc.2014.05.034.
[49] Vaidya, A.; Jain, S.; Jain, S.; Jain, A. K.; Agrawal, R. K. Journal of Pharmaceutical
Sciences and Pharmacology, 2014, 1, 219–232.
[50] Viswanathan, V. M.; Goshe, A. K.; Revanker, G. R.; Robbins, R. K. J. Chem. Inf.
Comput. Sci. 1989, 29, 163-172.
[51] Walters, W. P.; Stahl, M. T.; Murcko, M. A. Drug.Discov. Today. 1998, 3(4), 160–
178.
[52] Wang, J.; Chu, P.; Chen, C.; Lin, J. Nucleic Acids Research, 2012, 40.
[53] Wavefunction, (2004) Inc. Spantan’04, Irvin, California, USA.
[54] Wildman, S. A.; Crippen, G. M. J Chem Inf Comput Sci. 1999, 39, 868-873.
[55] Yap, C.W. J. Computational Chem. 2011, 37(7), 1466-1474

Copyright © 2020 The Author(s). This is an Open Access article distributed under the terms
of Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-
NC-ND 4.0), which permits anyone to share, use, reproduce and redistribute in any medium,
provided the original author and source are credited.

89