- Includes numerous practical examples related to QSAR methods and applications
- Follows the Organization for Economic Co-operation and Development principles for QSAR model development
- Discusses related techniques such as structure-based design and the combination of structure- and ligand-based design tools
Understanding the Basics of QSAR for Applications in Pharmaceutical Sciences and Risk Assessment describes the historical evolution of quantitative structure-activity relationship (QSAR) approaches and their fundamental principles. This book includes clear, introductory coverage of the statistical methods applied in QSAR and new QSAR techniques, such as HQSAR and G-QSAR. Containing real-world examples that illustrate important methodologies, this book identifies QSAR as a valuable tool for many different applications, including drug discovery, predictive toxicology and risk assessment. Written in a straightforward and engaging manner, this is the ideal resource for all those looking for general and practical knowledge of QSAR methods. Includes numerous practical examples related to QSAR methods and applications Follows the Organization for Economic Co-operation and Development principles for QSAR model development Discusses related techniques such as structure-based design and the combination of structure- and ligand-based design tools
Front Cover 1
Understanding the Basics of QSAR for Applications in Pharmaceutical Sciences and Risk Assessment 4
Copyright Page 5
Dedication 6
Contents 8
Foreword 12
Preface 14
1 Background of QSAR and Historical Developments 16
1.1 Introduction 17
1.2 Physicochemical Aspects of Biological Activity of Drugs and Chemicals 21
1.2.1 Hydrophobicity 21
1.2.2 Electronic effect 21
1.2.3 Steric effect 22
1.2.4 Forces and chemical bonding 22
1.2.4.1 Covalent bond 22
1.2.4.2 Ionic bond 23
1.2.4.3 Hydrogen bond 23
1.2.4.4 Hydrophobic force 25
1.2.4.5 van der Waals interaction 26
1.2.4.6 Pi–pi (p–p) stacking interaction 27
1.2.4.7 Charge transfer complex 29
1.2.4.8 Orbital-overlapping interaction 31
1.2.4.9 Ion-dipole and ion-induced dipole interaction 31
1.2.5 Structural features influencing response of chemicals 31
1.2.5.1 Stereochemical features influencing drug activity 32
1.2.5.2 Isosterism features influencing drug activity 32
1.2.5.3 Miscellaneous contribution of structural features 34
1.3 Structure–Activity Relationship 35
1.3.1 Ideology 35
1.3.2 The components and principal steps involved 38
1.3.3 Naming of the components 41
1.3.4 Objectives of QSAR model development 43
1.3.4.1 Prediction of activity/property/toxicity 43
1.3.4.2 Reduction and replacement of experimental (laboratory) animals 44
1.3.4.3 Virtual screening of library data 44
1.3.4.4 Diagnosis of mechanism 45
1.3.4.5 Classification of data 45
1.3.4.6 Optimization of leads 45
1.3.4.7 Refinement of synthetic targets 46
1.4 Historical Development of QSARs: A Journey of Knowledge Enrichment 46
1.5 Applications of QSAR 50
1.6 Regulatory Perspectives of QSAR 53
1.7 Overview and Conclusion 54
References 58
2 Chemical Information and Descriptors 62
2.1 Introduction 63
2.2 Concept of Descriptors 63
2.3 Type of Descriptors 68
2.3.1 Substituent constants 68
2.3.2 Whole molecular descriptors 69
2.4 Descriptors Commonly Used in QSAR Studies 69
2.4.1 Physicochemical descriptors 69
2.4.1.1 Hydrophobic parameters 69
2.4.1.1.1 Partition coefficient (log P) 69
2.4.1.1.2 Hydrophobic substituent constant (p) 70
2.4.1.1.3 Hydrophobic fragmental constant (f, f') 71
2.4.1.2 Electronic parameters 71
2.4.1.2.1 Acid dissociation constant 71
2.4.1.2.2 Hammett constant 72
2.4.1.3 Steric parameters 73
2.4.1.3.1 Taft steric constant 73
2.4.1.3.2 Charton’s steric parameter (.) and van der Waals radius 73
2.4.1.3.3 Effective Charton’s steric parameter (.ef) 74
2.4.1.3.4 STERIMOL parameters 74
2.4.1.3.5 Molar refractivity 75
2.4.1.3.6 Parachor 76
2.4.2 Topological descriptors 76
2.4.3 Structural descriptors 76
2.4.4 Indicator variables 76
2.4.5 Thermodynamic descriptors 82
2.4.6 Electronic parameters 83
2.4.7 Quantum chemical descriptors 84
2.4.7.1 Mulliken atomic charges 84
2.4.7.2 Quantum topological molecular similarity indices 84
2.4.8 Spatial parameters 84
2.4.8.1 RadofGyration 85
2.4.8.2 Jurs descriptors 85
2.4.8.3 Shadow indices 86
2.4.8.4 Molecular surface area 86
2.4.8.5 Density 87
2.4.8.6 Principal moment of inertia 88
2.4.8.7 Molecular volume 88
2.4.9 Information indices 88
2.4.9.1 Information of atomic composition index 89
2.4.9.2 Information indices based on the A-matrix 89
2.4.9.3 Information indices based on the D-matrix 89
2.4.9.4 Information indices based on the E-matrix and the ED-matrix 90
2.4.9.5 Multigraph information content indices (IC, BIC, CIC, SIC) 90
2.4.10 Molecular shape analysis descriptors 90
2.4.11 Molecular field analysis parameters 91
2.4.12 Receptor surface analysis parameters 91
2.5 Overview and Conclusion 93
References 93
3 Classical QSAR 96
3.1 Introduction 96
3.2 The Free–Wilson Model 97
3.2.1 The concept 97
3.2.2 The methodology 97
3.2.3 Example of Free–Wilson model development 98
3.3 The Fujita–Ban Model 103
3.3.1 The concept 103
3.3.2 The methodology 104
3.4 The LFER Model 107
3.4.1 The concept 107
3.4.2 Genesis 107
3.4.3 An example 111
3.4.4 Applications 113
3.5 Kubinyi’s Bilinear Model 113
3.6 The Mixed Approach 115
3.7 Overview and Conclusions 116
References 116
4 Topological QSAR 118
4.1 Introduction 118
4.2 Topology: A Method of Chemical Structure Representation 119
4.3 Graphs and Matrices: Platforms for the Topological Paradigm 120
4.3.1 Graph theory and chemical graphs 120
4.3.2 Matrix: aiding the numerical presentation of graph theory 123
4.4 Topological Indices 136
4.4.1 The context and formalism 136
4.4.2 Wiener, Platt, Hosoya, Zagreb, and Balaban indices 137
4.4.2.1 Wiener index (W) 137
4.4.2.2 Platt number (F) 138
4.4.2.3 Hosoya index (Z) 138
4.4.2.4 Zagreb index 138
4.4.2.5 Balaban index (J) 140
4.4.3 Molecular connectivity indices 140
4.4.3.1 Randic connectivity index 141
4.4.3.2 Kier and Hall’s connectivity index 142
4.4.4 Kappa shape indices 146
4.4.5 Electrotopological state (E-state) indices 152
4.4.6 Extended topochemical atom indices 154
4.5 Conclusion and Possibilities 155
References 162
5 Computational Chemistry 166
5.1 Introduction 166
5.2 Computer Use in Chemistry 167
5.2.1 Visualization 168
5.2.1.1 Structure drawing 168
5.2.1.2 3D visualization 169
5.2.1.3 Visualization of ligand–receptor interactions 170
5.2.2 Calculation and simulation 170
5.2.3 Analysis and storage of data 171
5.3 Conformational Analysis and Energy Minimization 173
5.3.1 The concept 173
5.3.2 Conformational search 175
5.3.3 Minimization of energy 175
5.4 Molecular Mechanics 180
5.5 Molecular Dynamics 185
5.5.1 Definition 185
5.5.2 Development and components 186
5.5.3 The algorithm 187
5.6 Quantum Mechanics 188
5.6.1 The Born–Oppenheimer approximation 194
5.6.2 The Hartree–Fock approximation 194
5.6.3 Density functional theory 195
5.6.4 Semiempirical analysis 196
5.6.4.1 Concept 196
5.6.4.2 Developmental background 197
5.6.4.3 Modified neglect of diatomic overlap 197
5.6.4.4 Austin model 1 199
5.6.4.5 Parametric method 3 199
5.6.4.6 PDDG/PM3 and PDDG/MNDO 200
5.7 Overview and Conclusion 200
References 203
6 Selected Statistical Methods in QSAR 206
6.1 Introduction 206
6.2 Regression-Based Approaches 207
6.2.1 Multiple linear regression 207
6.2.1.1 Model development 207
6.2.1.2 Statistical metrics to examine the quality of the developed model 211
6.2.1.2.1 Mean average error 211
6.2.1.2.2 Determination coefficient (R2) 211
6.2.1.2.3 Adjusted R2 (Ra2) 212
6.2.1.2.4 Variance ratio (F) 212
6.2.1.2.5 Standard error of estimate (s) 213
6.2.1.2.6 Root mean square error of calibration 213
6.2.1.2.7 The “t” test for each regression coefficient 213
6.2.1.2.8 Intercorrelation among descriptors 214
6.2.1.3 Example of an MLR model development 214
6.2.1.4 Data pretreatment and variable selection 222
6.2.1.4.1 Data set curation 222
6.2.1.4.2 Data pretreatment 223
6.2.1.4.3 Variable selection (feature selection) 223
6.2.1.4.3.1 Stepwise selection 223
6.2.1.4.3.2 All possible subset selection 224
6.2.1.4.3.3 Genetic method 224
6.2.1.4.3.4 Factor analysis 225
6.2.1.4.3.5 Other methods 225
Particle swarm optimization 225
Ant colony optimization 226
k-Nearest neighborhood method 226
6.2.2 Partial least squares 226
6.2.2.1 The method 227
6.2.2.2 An example 228
6.2.3 Principal component regression analysis 229
6.2.4 Ridge regression 230
6.3 Classification-Based QSAR 230
6.3.1 Linear discriminant analysis 231
6.3.2 Logistic regression 231
6.3.3 Cluster analysis 233
6.3.3.1 Hierarchical cluster analysis 234
6.3.3.2 k-Means clustering 234
6.4 Machine Learning Techniques 235
6.4.1 Artificial neural network 235
6.4.2 Bayesian neural network 240
6.4.3 Decision tree and random forest 241
6.4.4 Support vector machine 242
6.5 Conclusion 243
References 243
7 Validation of QSAR Models 246
7.1 Introduction 246
7.2 Different Validation Methods 248
7.2.1 The OECD principles 249
7.2.2 Internal validation 251
7.2.3 External validation 251
7.2.3.1 Division of the data set into training and test sets 251
7.2.3.2 Applicability domain 254
7.2.3.2.1 Concept of the AD 254
7.2.3.2.2 History behind the introduction of the AD 254
7.2.3.2.3 Types of AD approaches 255
7.2.3.2.4 Checklist and importance of the AD study in validation 268
7.2.4 Validation metrics 269
7.2.4.1 Validation metrics for regression-based QSAR models 269
7.2.4.1.1 Metrics for internal validation 269
7.2.4.1.1.1 Leave-one-out cross-validation 269
7.2.4.1.1.2 Leave-many-out cross-validation 273
7.2.4.1.1.3 True Q2 273
7.2.4.1.1.4 The r2m metric for internal validation 274
7.2.4.1.1.5 True r2m(LOO) 277
7.2.4.1.1.6 Bootstrapping 277
7.2.4.1.1.7 Metrics for chance correlation: Y-randomization 277
7.2.4.1.2 Metrics for external validation 278
7.2.4.1.2.1 Predictive R2 (R2pred) 278
7.2.4.1.2.2 Validation based on Golbraikh and Tropsha’s criteria 278
7.2.4.1.2.3 The r2m(test) metric for external validation 279
7.2.4.1.2.4 RMSEP 280
7.2.4.1.2.5 Q2(F2) 280
7.2.4.1.2.6 Q2(F3) 280
7.2.4.1.2.7 Concordance correlation coefficient 281
7.2.4.1.2.8 The r2m(rank) metric 281
7.2.4.2 Validation metrics for classification-based QSAR models 282
7.2.4.2.1 Goodness-of-fit and quality measures 282
7.2.4.2.1.1 Wilks lambda (.) statistics 282
7.2.4.2.1.2 Canonical index (Rc) 283
7.2.4.2.1.3 Chi-square (.2) 283
7.2.4.2.1.4 Squared Mahalanobis distance 283
7.2.4.2.2 Metrics for model performance parameters 283
7.2.4.2.2.1 Sensitivity, specificity, and accuracy 283
7.2.4.2.2.2 F-measure and precision 284
7.2.4.2.2.3 G-means 284
7.2.4.2.2.4 Cohen’s . 285
7.2.4.2.2.5 Matthews correlation coefficient 285
7.2.4.2.3 Parameters for receiver operating characteristics analysis 286
7.2.4.2.3.1 ROC curve 286
7.2.4.2.3.2 ROCED and ROCFIT 287
7.2.4.2.3.3 AUC-ROC 288
7.2.4.2.4 Metrics for Pharmacological distribution diagram 288
7.3 A Practical Example of the Calculation of Common Validation Metrics and the AD 289
7.4 QSAR model reporting format 299
7.4.1 Concept of the QMRF 299
7.4.2 Why QMRF? 299
7.4.3 How to construct QMRF 300
7.4.4 Utility of the QMRF 300
7.5 Overview and Conclusion 300
References 301
8 Introduction to 3D-QSAR 306
8.1 Introduction 307
8.2 Comparative Molecular Field Analysis 308
8.2.1 Concept of CoMFA 308
8.2.2 Methodology of CoMFA 309
8.2.3 Factors responsible for the performance of CoMFA 310
8.2.3.1 Biological data 310
8.2.3.2 Optimization of 3D structure of the compounds 311
8.2.3.3 Conformational analysis of compounds 312
8.2.3.4 Determination of bioactive conformations 313
8.2.3.4.1 X-ray crystallography 313
8.2.3.4.2 NMR spectroscopy 314
8.2.3.5 Alignment of molecules 314
8.2.3.6 Calculation of molecular interaction energy fields 315
8.2.3.7 Model generation 315
8.2.4 Display and interpretation of results 316
8.2.5 Advantages and drawbacks of CoMFA 316
8.3 Comparative Molecular Similarity Indices Analysis 317
8.3.1 Concept of comparative molecular similarity indices analysis 317
8.3.2 Methodology of CoMSIA 317
8.3.3 Advantages of CoMSIA 318
8.4 Molecular Shape Analysis 319
8.4.1 Concept of molecular shape analysis 319
8.4.2 Methodology of the MSA 320
8.4.3 MSA descriptors 321
8.5 Receptor Surface Analysis 322
8.5.1 Concept of receptor surface analysis 322
8.5.2 Methodology of the RSA 322
8.5.3 RSA descriptors 322
8.6 Other Approaches 323
8.6.1 Alignment-based 3D-QSAR models 323
8.6.1.1 Self-organizing molecular field analysis 323
8.6.1.2 Voronoi field analysis 324
8.6.1.3 Molecular quantum similarity measures 325
8.6.1.4 Adaptation of the fields for molecular comparison 325
8.6.1.5 Genetically evolved receptor modeling 326
8.6.1.6 Hint interaction field analysis 327
8.6.2 Alignment-independent 3D-QSAR models 327
8.6.2.1 Comparative molecular moment analysis 327
8.6.2.2 Weighted holistic invariant molecular descriptor analysis 328
8.6.2.3 VolSurf 328
8.6.2.4 Compass 328
8.6.2.5 GRID 329
8.6.2.6 Comparative spectral analysis 330
8.6.2.7 Quantum chemical parameters in QSAR analysis 330
8.7 Overview and Conclusions 330
References 331
9 Newer QSAR Techniques 334
9.1 Introduction 335
9.2 HQSAR 336
9.2.1 Concept of HQSAR 336
9.2.2 How to develop an HQSAR model 336
9.2.3 HQSAR parameters 338
9.2.3.1 Hologram length 338
9.2.3.2 Fragment size 339
9.2.3.3 Fragment distinction 339
9.2.4 Why use HQSAR over other techniques? 339
9.2.5 Application of HQSAR models 340
9.2.5.1 A flexible tool in drug design 340
9.2.5.2 Mathematical correlation to activity/property prediction 342
9.2.5.3 Pharmacokinetic studies and ADME prediction 343
9.3 G-QSAR 344
9.3.1 Concept of G-QSAR 344
9.3.2 Background of evaluation of G-QSAR method 344
9.3.3 G-QSAR methodology 345
9.3.3.1 Molecular fragmentation 346
9.3.3.2 Calculation of fragment descriptors 347
9.3.3.3 G-QSAR model development 347
9.3.4 Application of the G-QSAR model 347
9.3.4.1 NCE design based on fragments 348
9.3.4.2 Scaffold hopping and lead optimization 349
9.3.4.3 Addressing the inverse QSAR problem 349
9.3.4.4 Mathematical correlation to activity prediction 350
9.4 Other Approaches 351
9.4.1 MIA-QSAR 351
9.4.1.1 Concept of MIA-QSAR 351
9.4.1.2 Methodology of MIA-QSAR 352
9.4.1.2.1 Descriptor calculation 352
9.4.1.2.2 Model development 352
9.4.1.3 Pros and cons of MIA-QSAR 353
9.4.1.4 Application of MIA-QSAR 354
9.4.2 Binary QSAR 355
9.4.2.1 Concept of binary QSAR 355
9.4.2.2 Methodology of binary QSAR 355
9.4.3 Fragment-based QSAR 356
9.4.4 Fragment-similarity-based QSAR 357
9.4.5 Ensemble QSAR 358
9.4.5.1 Concept of eQSAR 358
9.4.5.2 Importance and application of eQSAR 359
9.4.6 LQTA-QSAR 359
9.4.6.1 Concept of LQTA-QSAR 359
9.4.6.2 Methodology of LQTA-QSAR 360
9.4.7 SOM 4D-QSAR 360
9.4.8 Receptor-independent 4D-QSAR 362
9.4.9 Receptor-dependent 4D-QSAR 364
9.4.10 5D-QSAR (QUASAR) 367
9.4.11 6D-QSAR 367
9.4.12 7D-QSAR 368
9.5 Overview and Conclusions 368
References 369
10 Other Related Techniques 372
10.1 Introduction 373
10.2 Pharmacophore 374
10.2.1 Concept and definition 374
10.2.2 Background and early days of pharmacophore 375
10.2.3 Methodology of pharmacophore mapping 376
10.2.3.1 Diverse conformation generation 376
10.2.3.2 Generation of 3D pharmacophore 377
10.2.3.3 Assessment of the quality of pharmacophore hypotheses 378
10.2.3.4 Validation of the pharmacophore model 380
10.2.4 Types of pharmacophore 381
10.2.4.1 Ligand-based pharmacophore modeling 381
10.2.4.2 Structure-based pharmacophore modeling 383
10.2.5 Application of pharmacophore models 387
10.2.5.1 Pharmacophore model–based VS 387
10.2.5.2 Pharmacophore-based de novo design 388
10.2.6 Advantages and limitations of pharmacophore 389
10.2.7 Software tools for pharmacophore analysis 390
10.3 Structure-Based Design–Docking 390
10.3.1 Concept and definition of docking 390
10.3.2 Definition of fundamental terms of docking 394
10.3.3 Essential requirements of docking 396
10.3.4 Categorization of docking 397
10.3.4.1 Receptor/protein flexibility 398
10.3.4.1.1 Soft docking 398
10.3.4.1.2 Side-chain flexibility 399
10.3.4.1.3 Molecular relaxation 400
10.3.4.1.4 Docking of multiple protein structures/ensemble docking 400
10.3.4.2 Ligand sampling and flexibility 400
10.3.4.2.1 Shape matching 401
10.3.4.2.2 Systematic search 401
10.3.4.2.3 Stochastic algorithms 402
10.3.4.3 Docking scoring functions 402
10.3.4.3.1 FF scoring functions 402
10.3.4.3.2 Empirical scoring functions 402
10.3.4.3.3 Knowledge-based scoring functions 403
10.3.4.3.4 Consensus scoring 403
10.3.4.3.5 Clustering- and entropy-based scoring methods 403
10.3.5 Basic steps of docking 403
10.3.6 Challenges and required improvements in docking studies 404
10.3.7 Applications of docking 407
10.3.8 Docking software tools 408
10.4 Combination of Structure- and Ligand-Based Design Tools 411
10.4.1 Comparative binding energy analysis 411
10.4.1.1 The concept of comparative binding energy 411
10.4.1.2 The methodology of COMBINE 412
10.4.1.3 Importance and advantages of COMBINE 414
10.4.1.4 Drawbacks and required improvements 414
10.4.1.5 Applications of COMBINE 414
10.4.1.6 Software for COMBINE 415
10.4.2 Comparative residue interaction analysis 415
10.4.2.1 Concept of CoRIA 415
10.4.2.2 Methodology of CoRIA 416
10.4.2.3 Variants of CoRIA 416
10.4.2.4 Importance and application of CoRIA 417
10.4.2.5 Drawback of CoRIA 417
10.4.2.6 Future perspective of CoRIA 418
10.5 In silico Screening of Chemical Libraries: VS 418
10.5.1 Concept 418
10.5.2 Workflow and types of VS 418
10.5.2.1 Selection of chemical libraries/databases 419
10.5.2.2 Preprocessing of chemical libraries 419
10.5.2.3 Filtering of druglike molecules 419
10.5.2.4 Screening 419
10.5.2.5 Hit selection to new chemical entity generation 420
10.5.3 Successful application of VS: A few case studies 421
10.5.4 Advantages of VS 421
10.5.4.1 Cost-effective 428
10.5.4.2 Time-saving 428
10.5.4.3 Labor-efficient 428
10.5.4.4 Sensible alternative 428
10.5.5 Pitfalls 428
10.5.6 Databases for the VS 431
10.6 Overview and Conclusions 432
References 436
11 SAR and QSAR in Drug Discovery and Chemical Design—Some Examples 442
11.1 Introduction 442
11.2 Successful Applications of QSAR and Other In Silico Methods: Representative Examples 443
11.2.1 Examples of some approved drugs 443
11.2.2 Examples of other approved chemicals 459
11.2.3 Examples of investigational drugs at different phases of current clinical trials 461
11.3 Conclusion 465
References 465
12 Future Avenues 470
12.1 Introduction 470
12.2 Application Areas 471
12.2.1 QSAR of mixture toxicity 471
12.2.2 Peptide QSAR 471
12.2.3 QSAR of nanoparticles 471
12.2.4 QSAR of ionic liquids 472
12.2.5 QSAR of cosmetics 473
12.2.6 PKPD-linked QSAR modeling 473
12.2.7 Material informatics 474
12.2.8 Ecotoxicity modeling of pharmaceuticals 474
12.2.9 Interspecies toxicity modeling 475
12.2.10 QSAR of phytochemicals 475
12.3 Conclusion 475
References 475
Index 478
Chemical Information and Descriptors
Computational modeling, quantitative structure–activity relationship (QSAR) in particular, plays an important role in the chemistry disciplines ranging from drug discovery to materials science. Numerical portrayal of molecular structures encoding the required chemical information responsible for a given molecular property (or activity) is the first step in a QSAR analysis. This numerical depiction of molecular structure information is carried out through computation of descriptors, which can be considered in many forms, from simple atom counts to complex molecular features. Molecular descriptors play a fundamental role in cheminformatics and chemometric analyses. The concept of chemical information and descriptors, as well as a broad categorization of descriptors, are highlighted in this chapter.
Keywords
Descriptors; graph theory; physicochemical; electronic; structural; topological; quantum
Contents
2.3.1 Substituent constants 53
2.3.2 Whole molecular descriptors 54
2.4 Descriptors Commonly Used in QSAR Studies 54
2.4.1 Physicochemical descriptors 54
2.4.1.1 Hydrophobic parameters 54
2.4.1.2 Electronic parameters 56
2.4.2 Topological descriptors 61
2.4.3 Structural descriptors 61
2.4.5 Thermodynamic descriptors 67
2.4.6 Electronic parameters 68
2.4.7 Quantum chemical descriptors 69
2.4.7.1 Mulliken atomic charges 69
2.4.7.2 Quantum topological molecular similarity indices 69
2.4.8.4 Molecular surface area 71
2.4.8.6 Principal moment of inertia 73
2.4.9.1 Information of atomic composition index 74
2.4.9.2 Information indices based on the A-matrix 74
2.4.9.3 Information indices based on the D-matrix 74
2.4.9.4 Information indices based on the E-matrix and the ED-matrix 75
2.4.9.5 Multigraph information content indices (IC, BIC, CIC, SIC) 75
2.4.10 Molecular shape analysis descriptors 75
2.4.11 Molecular field analysis parameters 76
2.4.12 Receptor surface analysis parameters 76
2.5 Overview and Conclusion 78
References 78
2.1 Introduction
The quantitative structure–activity relationship (QSAR) technique, being directly related to the molecular structures of chemicals, can explain the effects exerted by the chemicals in relation to their structures and properties. Any significant search for the required chemical information of molecules for a particular end point can provide a strong tool for the predictive assessment of the response of existing untested as well as new chemicals [1]. QSAR is a simple mathematical model that can correlate chemistry with the properties (physicochemical/biological/toxicological) of molecules using various computationally or experimentally derived quantitative parameters known as descriptors. These descriptors are correlated with the response variable using a variety of chemometric tools in order to obtain a meaningful QSAR model. The developed models provide a significant insight regarding the essential structural requisites of the molecules, thus enabling us to identify the features contributing to the biological activity/property/toxicity of the studied molecules [2].
2.2 Concept of Descriptors
Molecular descriptors are terms that characterize specific information about a studied molecule. They are the “numerical values associated with the chemical constitution for correlation of chemical structure with various physical properties, chemical reactivity, or biological activity” [3,4]. In other words, the modeled response (activity/property/toxicity of query molecules) is represented as a function of quantitative values of structural features or properties that are termed as descriptors for a QSAR model. Cheminformatics methods depend on the generation of chemical reference spaces into which new chemical entities are predictable by the developed QSAR model. The definition of chemical spaces significantly depends on the use of computational descriptors of studied molecular structure, physical or chemical properties, or specific features.
(activity/property/toxicity)=f(Information in form of chemical structureor property)=f(Descriptors)
The type of descriptors used and the extent to which they can encode the structural features of the molecules that are correlated to the response are critical determinants of the quality of any QSAR model. The descriptors may be physicochemical (hydrophobic, steric, or electronic), structural (based on frequency of occurrence of a substructure), topological, electronic (based on molecular orbital calculations), geometric (based on a molecular surface area calculation), or simple indicator parameters (dummy variables). A schematic overview is presented in Figure 2.1 in order to show the steps how a chemical structure is used to calculate descriptors and used in QSAR model development.
Figure 2.1 How chemical structure is used to calculate descriptors and QSAR model development.
A dimension in the QSAR analysis acts as the constraint that controls the nature of the analysis. The term dimension in predictive model development is roughly associated with the complexity of the modeling technique that directly signifies the degree of descriptors. The dimension of an object can be mathematically attributed to the minimum number of coordinates needed for specifying a particular point in it [1]. The addition of dimension to a specific geometric object assists in identifying it in a different way by adding more information. Thus, it is clear that dimension is an intrinsic property of an object and does not depend on the space of the object [1]. The addition of new dimensions to the QSAR technique helps in deriving structural information at a higher level of analysis. With the use of ascending dimensions of descriptors in the modern QSAR analysis, a QSAR modeler may be able to reveal new features of the molecules. The dimensionality of descriptors depends on the type of algorithm employed and defines the nature of QSAR analysis. In the development of a predictive model, the dimension is assigned on the basis of the nature of the independent variables (descriptors) and the corresponding QSAR modeling is named likewise; that is, a QSAR model comprising of one-dimensional (1D) parameters is called 1D-QSAR. In other words, one can conclude that the dimension of the performed QSAR analysis follows the dimension of the descriptor.
In order to pursue a quantitative analysis on structure of chemical compounds, generation of data encoding chemical information is an essential first step in the development of the QSAR model. It is therefore envisaged that QSAR analysis attempts to develop predictive models in the form of mathematical relations by using chemical information about molecules. Descriptors represent the chemical information that encodes the behavior of a molecular entity. They are the numerical or quantitative representations of chemical compounds derived using suitable algorithms and are used as independent variables for predictive model development. In summary, any apt structural information quantitatively describing the biological activity/property/toxicity of a molecule can be defined as a descriptor. Hence, molecular descriptors range from simple atomic counts or molecular weight measures to complex spatial or geometrical features [5].
One can describe a single molecule in many ways. It is possible to compute thousands of numerical descriptors for a given chemical. Many of these descriptors are very closely related to each other and even capture the same information at times. Thus, the selection of relevant descriptors is a well-known problem, and it requires a lot of experience for the QSAR modeler to select the appropriate ones for the model development [6]. In addition, one has to take into account the nature of the chemical structure being considered. A set of descriptors may efficiently encode the chemical information perfectly for the small molecules, but the same set of descriptors may not be able to encode the required features for polymers, protein structures, and inorganic molecules. Thus, not only the calculation...
| Erscheint lt. Verlag | 3.3.2015 |
|---|---|
| Sprache | englisch |
| Themenwelt | Medizin / Pharmazie ► Gesundheitsfachberufe |
| Medizin / Pharmazie ► Medizinische Fachgebiete ► Pharmakologie / Pharmakotherapie | |
| ISBN-10 | 0-12-801633-7 / 0128016337 |
| ISBN-13 | 978-0-12-801633-6 / 9780128016336 |
| Haben Sie eine Frage zum Produkt? |
PDF (Adobe DRM)Größe: 28,1 MB
Kopierschutz: Adobe-DRM
Adobe-DRM ist ein Kopierschutz, der das eBook vor Mißbrauch schützen soll. Dabei wird das eBook bereits beim Download auf Ihre persönliche Adobe-ID autorisiert. Lesen können Sie das eBook dann nur auf den Geräten, welche ebenfalls auf Ihre Adobe-ID registriert sind.
Details zum Adobe-DRM
Dateiformat: PDF (Portable Document Format)
Mit einem festen Seitenlayout eignet sich die PDF besonders für Fachbücher mit Spalten, Tabellen und Abbildungen. Eine PDF kann auf fast allen Geräten angezeigt werden, ist aber für kleine Displays (Smartphone, eReader) nur eingeschränkt geeignet.
Systemvoraussetzungen:
PC/Mac: Mit einem PC oder Mac können Sie dieses eBook lesen. Sie benötigen eine
eReader: Dieses eBook kann mit (fast) allen eBook-Readern gelesen werden. Mit dem amazon-Kindle ist es aber nicht kompatibel.
Smartphone/Tablet: Egal ob Apple oder Android, dieses eBook können Sie lesen. Sie benötigen eine
Geräteliste und zusätzliche Hinweise
Zusätzliches Feature: Online Lesen
Dieses eBook können Sie zusätzlich zum Download auch online im Webbrowser lesen.
Buying eBooks from abroad
For tax law reasons we can sell eBooks just within Germany and Switzerland. Regrettably we cannot fulfill eBook-orders from other countries.
EPUB (Adobe DRM)Größe: 23,0 MB
Kopierschutz: Adobe-DRM
Adobe-DRM ist ein Kopierschutz, der das eBook vor Mißbrauch schützen soll. Dabei wird das eBook bereits beim Download auf Ihre persönliche Adobe-ID autorisiert. Lesen können Sie das eBook dann nur auf den Geräten, welche ebenfalls auf Ihre Adobe-ID registriert sind.
Details zum Adobe-DRM
Dateiformat: EPUB (Electronic Publication)
EPUB ist ein offener Standard für eBooks und eignet sich besonders zur Darstellung von Belletristik und Sachbüchern. Der Fließtext wird dynamisch an die Display- und Schriftgröße angepasst. Auch für mobile Lesegeräte ist EPUB daher gut geeignet.
Systemvoraussetzungen:
PC/Mac: Mit einem PC oder Mac können Sie dieses eBook lesen. Sie benötigen eine
eReader: Dieses eBook kann mit (fast) allen eBook-Readern gelesen werden. Mit dem amazon-Kindle ist es aber nicht kompatibel.
Smartphone/Tablet: Egal ob Apple oder Android, dieses eBook können Sie lesen. Sie benötigen eine
Geräteliste und zusätzliche Hinweise
Zusätzliches Feature: Online Lesen
Dieses eBook können Sie zusätzlich zum Download auch online im Webbrowser lesen.
Buying eBooks from abroad
For tax law reasons we can sell eBooks just within Germany and Switzerland. Regrettably we cannot fulfill eBook-orders from other countries.
aus dem Bereich
