QSAR STUDIES & DESIGNING OF POTENT HETEROCYCLIC COMPOUNDS AS γ-SECRETASE INHIBITORS

QSAR STUDIES & DESIGNING OF POTENT HETEROCYCLIC COMPOUNDS AS γ-SECRETASE INHIBITORS

Jagdeesh Ahirwar*, A. B. Mundada and A. K. Pathak

Department of Pharmacy, Barkatullah University, Bhopal-462026, Madhya Pradesh, India

ABSTRACT

Quantitative structure activity relationship (QSAR) studies ware performed on series of structurally similar heterocyclic sulfonamide with enzyme gamma secretase inhibitor activity. The compounds were divided into training and test set and generated different QSAR models using V-Life MDS 3.5 software multiple linear regression (MLR) method. Best QSAR models were selected on the basis of various statistical parameters like square correlation coefficient (r2), cross validated square correlation coefficient (q2), pred_r2, standard error of estimation (SE) and sequential Failure test (F). 2D QSAR study reveals that gamma secretase inhibitor activity is governed by physicochemical Alignment Independent (AI) descriptors and design new compounds, with more potent activity. The best models were found to be Model–I model-II and model-III. Model-I having 5 descriptors, r² ₌ 0.8582, q² ₌ 0.5701, Failure test = 22.9812 and predicted r² ₌ 0.7513. Model-II having 4 descriptors, r² ₌ 0.8170, q² ₌ 0.6780, Failure test = 18.9765 and predicted r² ₌ 0.6193. Model-III having 4 descriptors, r² ₌ 0.8248, q² ₌ 0.7006, Failure test = 20.0027 and predicted r² ₌ 0.7791. On the basis of 2D descriptors we designed many compounds in which compound J49 have the highest potency (EC₅₀ = 0.010 nM) in the design molecules as well as reported series.

MLR

INTRODUCTION: Alzheimer’s disease (AD) is a progressive neurodegenerative disease characterized pathologically neuronal loss, synaptic damage, loss of cholinergic activity in susceptible brain regions and abnormal deposits of insoluble extracellular plaques composed of β-amyloid peptides and intracellular neurofibrillary tangles composed of hyper-phosphorylated tau protein in the brain ^1-11.

Aβ peptides are released by sequential cleavage of the amyloid precursor protein (APP) in the amyloidogenic pathway by the action of two proteolytic membrane associated aspartic proteases. Firstly, β-secretase (BACE; β-amyloid precursor protein converting enzyme) cleaves amyloid precursor protein (APP) to form β-C-terminal fragment (β-CTF) and then γ-secretase cleaves β-CTF to form Aβ (Aβ₄₀ and Aβ₄₂) peptides and the cytosolic amyloid precursor protein intracellular domain (AICD) ^12-17.

γ-secretase is a membrane-embedded multiprotein complex consisting of at least four components; the Presenilin (PS) heterodimer, Nicastrin, Anterior Pharynx Defective (APH-1) and Presenilin Enhancer-2 (PEN-2) ¹⁸. Another reasons for Alzheimer disease includes destruction of cholinergic neuron due to oxidative stress and or decrease in level of choline acetyl transferase ¹⁹, genetic mutations.

APP is cleaved by α or β-secretase followed by γ-secretase to release P3 or Aβ peptides, respectively. Since β-secretase and γ-secretase are responsible for the production of amyloid peptide (Aβ), which is believed to play a central role in the neuropathology of Alzheimer’s disease ^{20, 21}. There is an urgent need to design γ-secretase inhibitors with higher bioactivities and also need to analyze the correlation between gamma secretase inhibitors activity and physico-chemical parameters of each category of compounds using the Quantitative Structure Activity Relationship (QSAR) methods because the quantitative analysis of such molecules can be utilized for increasing the potency and minimizing the side effects.

MATERIALS AND METHODS: QSAR studies and designing were performed on a computer using the software VLifeMDS 3.5 (V-life sciences technology Pvt. Ltd. Pune, India). For the QSAR studies a series of 47 compounds based heterocyclic sulfonamide was selected in which 36 compounds having definite γ-secretase inhibitory activity ¹. The biological activity log (1/EC₅₀ or –log EC₅₀) was calculated, subsequently used as dependent variable for the QSAR studies. The structure of the compounds with new code and biological activity was shown in table 1 (1a, 1b and 1c).

TABLE 1: LIST OF COMPOUNDS USED FOR THE QSAR STUDIES OF γ-SECRETASE INHIBITOR ACTIVITY

TABLE 1a

The * compound code indicate the potent where as ** on the compound code indicates the worst activity.

Table 1b

Table 1c

 Draw the 2D and 3D structure: The structure of molecule was drawn in 2D orientation with the help of software ACDLABS chemsketch 12.0 version and save in .mol2 file format. All 2D structures were converted into 3D orientation with the help of software VLifeMDS 3.5 version and save in .mol2 file format.

Energy minimization: Energy minimization ²² is the process of changing the geometry of a structure to reduce its energy. Lower the energy states are of interest because molecules preferentially adopt them. Consequently, they are more indicative of molecular behavior than their high energy neighbors. In many situations it is necessary to know about the thermodynamic properties like enthalpies, entropy, free energy and forces between atoms. The energies and optimized geometries of a molecule can calculate by using different force field and quantum mechanical methods. The force field method is to provide information about molecular structure, interaction between atoms in a molecule. The mathematical formulation of a typical molecular mechanics force field which also called the Potential Energy Function (PEF)²³, the potential energy function is a sum of many individual contributions (energy). It can be divided into bonded (Bond stretching, Angle bending, Torsions, Inversion (out of plane bending)) and non-bonded (Electrostatics, Van der waals forces, Hydrogen bonding) contribution, responsible for intramolecular and intermolecular interaction between atoms. There are many force field methods to calculate the energies and optimized geometries of molecule like universal force field (UFF) ^{24, 25}, Merck molecular force field (MMFF) ^26-29.

Selection of Training and Test: The biological data were divided in training and test set. Selection of molecules in the training set and test is a key and important feature of QSAR model. Therefore the care was taken in such a way that biological activities of all compounds in test set lie within the maximum and minimum value range of biological activities of training set of compounds. Generally Manual data selection method used for the selection of training and test.

Variable Selection Method:

Stepwise Forward Backward Method: In stepwise procedure a variable that entered the model in the earlier stages of selection may be deleted at the later stages. The calculations made for inclusion and elimination of variables are the same as forward selection and backward procedures. That is the stepwise method is essentially a forward selection procedure, but at each stage the possibility of deleting a variable, as in backward elimination, is considered. The number of variables retained in the model is based on the levels of significance assumed for inclusion and exclusion of variables from the model ³⁰.

Statistical Method: Regression methods are used to build a QSAR model in the form of a mathematical equation. This equation explains variation of one or more dependent variables (usually activity) in terms of independent variables (descriptors). The regression method (MLR) is the QSAR molecular models that were used to predict and design a compound with best possible inhibitory property.

• Multiple Linear Regressions: Multiple linear regressions (MLR) are the standard method for multivariate data analysis. This method of regression estimates the values of the regression coefficients by applying least squares curve fitting method. For getting reliable results, dataset having typically 5 times as many data points (molecules) as independent variables (descriptors) is required. The regression equation takes the form-

Y = B₁X₁ + B₂X₂ + B₃X₃ + C

Where ‘Y’ is the dependent variable, the ‘B’ is regression coefficients for corresponding ‘X’ (independent variable), ‘C’ is a regression constant or intercept.

The resulted MLR equations could describe the structure activity relationships well. However, due to the co-linearity problem in MLR analysis, we removed the collinear descriptors before MLR model development. Therefore, some information was discarded in MLR analysis ^31-37.

• k-Nearest Neighbor Molecular Field Analysis: 3D QSAR methods, k-nearest neighbor Molecular Field Analysis (k-NN MFA) ^38-41 requires suitable alignment of set of molecules. This is followed by generation of a common rectangular grid around the molecules. The steric and electrostatic energies are computed at the lattice points of the grid using methyl probe of charge +1. These interaction energy values at the grid points are considered for relationship generation using k-NN method and utilized as descriptors for obtaining distances within this method. An optimal training and test set can be generated for k-NN method using sphere exclusion method. This algorithm allows constructing training sets covering all descriptor space areas occupied by representative points. It is expected that the predictive ability of QSAR models generally decreases when the dissimilarity level increases. Once the training and test sets are generated, kNN methodology is applied to descriptors generated over grid.
• k-NN MFA with Stepwise Variable Selection Method: This method employs the k-NN classification principle combined with the stepwise variable selection procedure for optimization of (i) The number of nearest neighbors (k) used to estimate the activity of each compound (ii) Selection of variables from the original pool of all molecular descriptors (steric and electrostatic field at the lattice points) that are used to calculate similarities between compounds (i.e. distances in number of variable-dimensional descriptor space).

RESULT AND DISCUSSION:

2D qsar:

Unicolumn Statistic:

TABLE 2: UNICOLUMN STATISTICS OF DIFFERENT MODELS OF γ-SECRETASE INHIBITOR

In all model the max of training is higher than the test set where as in case of min. the test have high value than the training set. In all models the std. deviation lies in between 0.31 to 0.83.

TABLE 3: CORRELATION MATRIX OF DIFFERENT STRUCTURAL DESCRIPTORS ARISES IN DIFFERENT MODELS

A:  SssNHE-index B: SaasN(Noxide)E-index C: T_2_N_8, D: T_C_C_6  E: T_C_O_10 F: T_N_S_5 G: T_N_S_7, H:T_N_Cl_8

From the observation table it was seen that few descriptors has strong correlation with each other which has been given in the shaded portion of the table. Descriptor present in the individual model does not show correlation more than 0.5 with each other.

Equations of the various models:

Model-1: PEC50 = 1.0360(± 0.1767) T_N_S_7 + 0.2370(± 0.0247) T_C_O_10 + 0.4611(± 0.1172) SssNHE-index + 1.0235(± 0.2572) T_N_S_5 + 0.4888(± 0.1865) SaasN(Noxide)E-index - 4.8224

Model-2: PEC50 = 0.7344(± 0.1186) T_2_N_8 + 0.6448(± 0.0955) SssNHE-index - 0.6067(± 0.1696) T_N_Cl_8 + 0.0461(± 0.0093) T_C_C_6 - 4.2055

Model-3: PEC50 = 0.5697(± 0.0861) SssNHE-index + 0.7168(± 0.1340) T_2_N_8 - 0.4370 (± 0.1400) T_N_Cl_8 + 0.5140(± 0.1773) SaasN(Noxide)E-index - 4.0313

Importance of Descriptor:

SssNHE-index: - Electrotopological state indices for number of –NH group connected with two single bonds.

SaasN(Noxide)E-index:- Electrotopological state indices for number of nitro-oxide group connected with two aromatic and one single bond.

T_N_S_7:- This is the count of number of Nitrogen atoms (single double or triple bonded) separated from any sulphur atom (single double or triple bonded) by 7 bonds in a molecule.

T_N_S_5:- This is the count of number of Nitrogen atoms (single double or triple bonded) separated from any sulphur atom (single double or triple bonded) by 5 bonds in a molecule.

T_N_Cl_8:- This is the count of number of Nitrogen atoms (single double or triple bonded) separated from any chlorine atom (single double or triple bonded) by 8 bonds in a molecule.

T_2_N_8:- This is the count of number of double bounded atoms (i.e. any double bonded atom, T_2) separated from Nitrogen atom by 8 bonds.

T_C_O_10:- This is the count of number of Carbon atoms (single double or triple bonded) separated from any Oxygen atom (single or double bonded) by 10 bonds distance in a molecule.

T_C_C_6:- This is the count of number of Carbon atoms (single double or triple bonded) separated from any Carbon atom (single or double bonded) by 6 bonds distance in a molecule.

TABLE 4: VALUES OF DIFFERENT PARAMETERS GENERATED IN MODEL DURING 2D QSAR

 TABLE 5: PREDICATED ACTIVITY OF MODEL-1, 2 AND 3 GENERATED DURING 2D QSAR OF γ-SECRETASE INHIBITORS

P: Predicted activity, R: Residual. Structure code in bold indicates test set.

Fitness plots (Actual vs. Predicted activity) and contribution chart of different models:

GRAPH 1: FITNESS PLOT OF MODEL-1

CHART 2: CONTRIBUTION CHART OF MODEL-1

GRAPH 2: FITNESS PLOT OF MODEL-2

CHART 2: CONTRIBUTION CHART OF MODEL-2

GRAPH 3: FITNESS PLOT OF MODEL-3

CHART 3: CONTRIBUTION CHART OF MODEL-3

Interpretation of 2D equations: There are many type of descriptor that are found in 2D QSAR but only 4 descriptors are found to be strongly correlated with the biological activity. The common descriptors in 2D QSAR models of heterocyclic sulfonamide are SssNHE-index, T_N_S_7, T_N_Cl_8 and T_2_N_8.

The remaining descriptors that arise in the equation are SaasN(Noxide)E-index, T_C_O_10, T_N_S_5 and T_C_C_6. It is clear from contribution chart of different 2D models that SssNHE-index, T_N_S_7, T_2_N_8 and T_N_Cl_8 has strong correlation than other descriptors.

The contribution of SssNHE-index in all models is positive means it has positive contribution with the activity. The values of SssNHE-index are 16.87%, 36.48% and 37.08% in model-1, 2 and 3 respectively. This descriptor provides importance of NH groups that are connected with two single bonds. The compound J14 and J15 was not possessing SssNHE-index in the structure and has most poor the biological activity. So increase the NH group or addition of NH group in the molecule will lead to increase in activity.

The contribution of T_N_S_7 descriptor was found to be 28.55% in model-1. The importance of this descriptor was found be presence of nitrogen atom that was separated from sulphur by 7 bonds. In complete series sulphur is present in the thiophene ring and nitrogen present on pyrazole ring. So these nitrogen and sulphur are separated by 7 bonds and contribution is positive that indicates the 7 bonds difference between nitrogen and sulphur is required for the biological activity. Hence, this descriptor (SssNHE-index and T_N_S_7) is used for the designing of the molecule.

3d QSAR:

Alignment of molecule: 30 molecules have been selected for the 3D QSAR studies since structure code J06 to J11 contains phenyl moiety instead of thiophene. Since for 3D QSAR common core is essential. Hence to perform the 3D QSAR 30 molecules has been selected.

From the figure 1 and 2, the alignment of thiophene core has been achieved this leads to calculation of descriptors (Electrostatic and steric) from the non-align area of the molecule. The data that was obtained after the alignment of 30 molecules was shown in table 6.

FIG. 1: ALIGNMENT OF THIOPHENE

FIG. 2: ALIGNMENT OF THIOPHENE-2-SULPHONAMIDE

TABLE 6: ALIGNMENT RESULTS OF 30 COMPOUNDS BY USING THIOPHENE AND THIOPHENE-2-SULPHONAMIDE AS Γ-SECRETASE INHIBITOR

3D model: From hundreds of models that have been the best model that observed is given above. The different descriptors that has arises are: S_1322 (-0.0565, -0.0192), S_1871 (-0.3631, -0.1947), S_1482 (-0.1567, -0.0463) and S_1704 (-0.0994, -0.0291)

TABLE 7: VALUES OF DIFFERENT PARAMETERS GENERATED IN 3D MODEL

FIG. 3: SHOW POINTS IN A MODEL (3D QSAR)

Designing of potent γ-secretase inhibitor: On the basis of 2D descriptor SssNHE-index the –NH group is required for the activity in new design compound (Table 8). According to positive contribution value of T_C_C_6 is required for activity; means addition of chain length is helpful for activity of new design compound. Other descriptors like T_N_S_7 and SaasN(Noxide)E-index are also responsible for the activity.

Activity of all designed compound was predicted and compared with respected to the predicted activity of the reported compounds. γ-secretase inhibitor of all designed compound was predicted using the best 2D model. The entire new designed compound shows the good to potent activity. Compound J49 show the most potent activity. All the data that was obtained after designing is shown in table 8.

Lead moiety used for designing of potent γ-secretase inhibitor

Table 8: Structure of design molecule and predicted activity

Compound in bold indicate most potent compound

CONCLUSIONS: γ-secretase is membrane-bound proteases that process amyloid precursor protein (APP). APP is cleaved by γ-secretase to release Aβ peptides. Since γ-secretase are responsible for the production of amyloid peptide (Aβ), which is believed to play a central role in the neuropathology of Alzheimer’s disease (AD). QSAR is a branch of computational chemistry that provides the knowledge structure properties of the molecule and helps to correlate the structural properties and the biological activity via linear regression.

In 2D QSAR studies of heterocyclic sulfonamide different models has been generated. The best three models were found to be Model-1 model-2 and model-3. Model-1 having 5 descriptors, r² ₌ 0.8582, q² ₌ 0.5701, Failure test = 22.9892 and predicted r² ₌ 0.7513. Another model-2 having 4 descriptors, r² ₌ 0.8248, q² ₌ 0.7006, Failure test = 20.0027 and predicted r² ₌ 0.7791. The descriptor that was generated shows that the NH group connected with two single bonds, should as high for good activity.

For designing of potent molecule as γ-secretase inhibitor the suggestion provided are same i.e. increasing NH group (from 2D QSAR) and steric properties (from 3D QSAR). On this basis we have design 18 compounds. For validation of model-2 was found in J49. For validation of model-1 was found in J51. In both structure J49 and J51 we have increase the NH group connected with two single bonds. Structure J49 has the highest potency {EC₅₀ = 0.000010 µM (0.010nM)} in the design molecule whereas reported series have (EC₅₀ = 0.13 µM).

ACKNOWLEDGEMENTS: It is a great pleasure for me to acknowledge to Head of the Department Dr. A. K. Pathak (Department of Pharmacy Barkatullah University Bhopal, M.P. India), for providing me an opportunity to work under whose kind control, the smooth completion of my project became possible. I would like to express my heartful gratitude to Mr. A. B. Mundada, for their precious suggestions.

REFERENCES:

1. Lynn R, Anthony FK, Suzan HA, Kevin PA, Thomas JC, Boyd LH, al. Synthesis and structure-activity relationship of a novel series of heterocyclic sulfonamide γ-secretase inhibitors. Bioorganic & Medicinal Chemistry. 2009; 17(13):4708-17.
2. Silverberg GD, Miller MC, Machan JT, Johanson CE, Caralopoulos IN, Pascale CL, al., Amyloid and Tau accumulate in the brains of aged hydrocephalic rats. Brain Research. 2010; 1317:286-96.
3. Moreno RD, Inestrosa NC, Culwell AR, Alvarez J. Sprouting and abnormal contacts of nonmedullated axons and deposition of extracellular material induced by the amyloid precursor protein (APP) and other protease inhibitors. Brain Research. 1996; 718:13-24.
4. Simón AM, Schiapparelli L, Salazar-Colocho P, Escribano L, Río JD, Frechill D, et al., Overexpression of wild-type human APP in mice causes cognitive deficits and pathological features unrelated to Aβ levels. Neurobiology of Disease. 2009; 33(3):369-78.
5. Bu G, Marzolo MP. Review: Lipoprotein receptors and cholesterol in APP trafficking and proteolytic processing, implications for Alzheimer’s disease. Seminars in Cell & Developmental Biology. 2009; 20(2):191-200.
6. Arias C, Quiroz-Baez R, Rojas E. Oxidative stress promotes JNK-dependent amyloidogenic processing of normally expressed human APP by differential modification of , β and γ-secretase expression. Neurochemistry International. 2009; 55(7):662-70.
7. Judes P. Apolipoprotein E represents a potent gene-based therapeutic target for the treatment of sporadic Alzheimer’s disease. Alzheimer’s & Dementia. 2008; 4(1) supplement-1:S91–S97.
8. Nabeshima T, Yamada K. Animal models of Alzheimer's disease and evaluation of anti-dementia drugs. Pharmacology & Therapeutics. 2000; 88(2):93-113.
9. Philip T, Michael F. A review of genome mutation and Alzheimer’s disease. 2007; 22(1):15-33.
10. Sommer B. Alzheimer’s disease and the amyloid cascade hypothesis: ten years on. Current Opinion in Pharmacology. 2002; 2(1):87-92.
11. Epis R, Gardoni F, Marcello E, Genazzani A, Canonico PL, Luca MD. Review: Searching for new animal models of Alzheimer’s disease. European Journal of Pharmacology. 2010; 626(1):57-63.
12. Holloway MK, Hunt P, McGaughey GB. Structure and Modeling in the Design of β- and γ-Secretase Drug Development Research. 2009; 70(2):70-93.
13. Kume H, Kametani F. Aβ11-40/42 production without γ-secretase ε-site cleavage. Biochemical and Biophysical Research Communications. 2006; 349(4):1356-60.
14. Li L, Zhang X, Yang D, Luo G, Chen S, Le W. Hypoxia increases A beta generation by altering β- and γ-cleavage of APP. Neurobiology of Aging. 2009; 30(7):1091-8.
15. Ryder J, Su Y, Liu F, Li B, Zhou Y, Ni B. Divergent roles of GSK3 and CDK5 in APP processing. Biochemical and Biophysical Research Communications. 2003; 312(4):922-9.
16. Martens GJM, Collin RWJ. The coding sequence of amyloid-β precursor protein APP contains a neural-specific promoter element. Brain Research. 2006; 1087:41-51.
17. Wolfe MS. γ-Secretase in biology and medicine. Seminars in Cell & Developmental Biology. 2009; 20(2):219-24.
18. Edbauer D, Winkler E, Regula JT, Pesold B, Steiner H, Haass C. Reconstitution of γ-secretase Nature Cell Biology. 2003; 5: 486-8.
19. Mohandas E, Rajmohan V, Raghunath B. Neurobiology of Alzheimer’s disease. Indian Journal of Psychiatry. 2009; 51(1):55-61.
20. Fraering PC. Structural and Functional Determinants of γ-secretase, an Intramembrane Protease Implicated in Alzheimer’s disease. Current Genomics. 2007; 8(8):531-49.
21. Xu M, Lai MT, Huang Q, Castro JL, Timothy H, Alan N, et al., γ-Secretase: characterization and implication for Alzheimer disease therapy. Neurobiology of Aging. 2002; 23(6):1023-30.
22. Williams DA, Lemke TL. Foye’s Principles of Medicinal Chemistry, 5^th edition, Lippincott Williams & Wilkins Baltimore; 2002, 81.
23. Halgren TA. Potential energy functions. Current Opinion in Structural Biology. 1995; 5(2):205-10.
24. Rappe AK, Casewit CJ, Colwell KS, Goddard III WA, Skiff WM. UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations. Journal of the American chemical society 1992; 114(25):10024-35.
25. Casewit CJ, Colwell KS, Rappe AK, Application of a universal force field to organic molecules. Journal of the American chemical society. 1992; 114(25):10035-46.
26. Halgren TA. Merck molecular force field-I. Basis, form, scope, parameterization, and performance of MMFF94 Journal of Computational Chemistry. 1996; 17(5-6):490-519.
27. Halgren TA. Merck molecular force field-II. MMFF94 van der Waals and electrostatic parameters for intermolecular interactions. Journal of Computational Chemistry. 1996; 17(5-6):520-552.
28. Halgren TA. Merck molecular force field-III. Molecular geometries and vibrational frequencies for MMFF94. Journal of Computational Chemistry. 1996; 17(5-6):553-586.
29. Halgren TA. Merck molecular force field-IV. conformational energies and geometries for MMFF94. Journal of Computational Chemistry 1996; 17(5-6):587-615.
30. Xu L, Zhang WJ. Comparison of different methods for variable selection. Analytica Chimica Acta. 2001; 446(1-2):477-83.
31. Tomera JF, Harakal C. Multiple Linear Regression Analysis of Blood Pressure, Hypertrophy, Calcium and Cadmium in Hypertensive and Non-hypertensive States. Food and Chemical Toxicology. 1997; 35(7):713-8.
32. Haus F, Boissel O, Junter GA. Multiple regression modelling of mineral base oil biodegradability based on their physical properties and overall chemical composition. Chemosphere. 2003; 50(7):939-48.
33. Ghasemi J, Saaidpour S. Quantitative structure-property relationship study of n-octanol-water partition coefficients of some of diverse drugs using multiple linear regression. Analytica Chimica Acta. 2007; 604(2):99-106.
34. Clouser DL, Jurs PC. The simulation of ¹³C nuclear magnetic resonance spectra of dibenzofurans using multiple linear regression analysis and neural networks. Analytica Chimica Acta. 1996; 321(2-3):127-35.
35. Gonzalez AG, Gonzalez AD. Advantages of target factor analysis against multiple linear regression methods for testing model equations in linear free energy relationships. Analytica Chimica Acta. 1995; 312(3):295-306.
36. Polak AG. Analysis of multiple linear regression algorithms used for respiratory mechanics monitoring during artificial ventilation. computer methods and programs in biomedicine. 2010; 99(3):219-314.
37. Deeb O, Hemmateenejad B, Jaber A, Garduno-Juarez R, Miri R. Effect of the electronic and physicochemical parameters on the carcinogenesis activity of some sulfa drugs using QSAR analysis based on genetic-MLR and genetic-PLS. Chemosphere. 2007; 67(11): 2122-30.
38. Adam J, Sebastiano S, Fabio R. A parallel network of modified 1-NN and k-NN classifiers-application to remote-sensing image classification. Pattern Recognition Letters. 1998; 19(1):57-62.
39. Sengur A. An expert system based on principal component analysis, artificial immune system and fuzzy k-NN for diagnosis of valvular heart diseases. Computers in Biology and Medicine. 2008; 38(3):329-38.
40. Hodge VJ, Lees KJ, Austin JL. A high performance k-NN approach using binary neural networks. Neural Networks 2004; 17(3):441-58.
41. Pradhan J, Sharma R, Goyal A. 3D-QSAR studies on 2-arylcarbonyl-3-trifluromrthylquinoxaline 1,4-diNoxide derivatives and their reduced analogues using k-NN MFA approach. International Journal of Pharma Research and Development-Online. 2010; 2(6):1-13.
    

    ---

    How to cite this article:

    Ahirwar J, Mundada AB and Pathak AK: QSAR Studies & Designing of Potent Heterocyclic Compounds as γ-Secretase Inhibitors. Int J Pharm Sci Res. 3(11); 4349-4362.

    ---