Neural Networks in QSAR and Drug Design (eBook)

Front Cover 1
Neural Networks in QSAR and Drug Design 4
Copyright Page 5
Contents 6
Contributors 10
Preface 12
Chapter 1. Strengths and Weaknesses of the Backpropagation Neural Network in QSAR and QSPR Studies 14
Abstract 14
Introduction 14
Standard BNN Algorithm 16
Designing the Model 19
Selection of the Best BNN Model 28
Comparison of the Performances of a BNN Model with those Obtained with other Approaches 31
Software Availability 33
Hybrid Systems with BNN 33
Conclusion 36
Annex: Artificial Neural Networks (ANNs) on Internet 37
References 37
Chapter 2. AUTOLOGP Versus Neural Network Estimation of n-Octanol/Water Partition Coefficients 60
Abstract 60
Introduction 60
Materials and Methods 62
Results and Discussion 66
Concluding Remarks 70
References 71
Chapter 3. Use of a Backpropagation Neural Network and Autocorrelation Descriptors for Predicting the Biodegradation of Organic Chemicals 78
Abstract 78
Introduction 78
Biodegradation Data 79
Molecular Descriptors 89
Statistics 91
Modeling Results 92
References 94
Chapter 4. Structure–Bell-Pepper Odor Relationships for Pyrazines and Pyridines Using Neural Networks 96
Abstract 96
Introduction 97
Materials and Methods 98
Results and Discussion 103
Conclusion 105
References 106
Chapter 5. A Neural Structure–Odor Threshold Model for Chemicals of Environmental and Industrial Concern 110
Abstract 110
Introduction 111
Materials and Methods 112
Results and Discussion 122
Concluding Remarks 127
References 128
Chapter 6. Adaptive Resonance Theory Based Neural Networks Explored for Pattern Recognition Analysis of QSAR Data 132
Abstract 132
Introduction 133
Neuro-Physiological Basis of ART 134
Taxonomy and State-of-the-Art 135
Theory of ART-2a and FuzzyART 137
Data Preprocessing by Complement Coding 141
Quantification or Qualification? 142
Case Study I: Classification of Rose Varieties from their Headspace Analysis 143
Case Study II: Optimal Selection of Aliphatic Substituents 151
Conclusions 160
References 160
Chapter 7. Multivariate Data Display Using Neural Networks 170
Abstract 170
Introduction 170
Methods 172
Results and Discussion 180
Conclusions 186
References 187
Chapter 8. Quantitative Structure–Activity Relationships of Nicotinic Agonists 190
Abstract 190
Introduction 191
Methods 194
Results 201
Discussion 209
References 217
Chapter 9. Evaluation of Molecular Surface Properties Using a Kohonen Neural Network 222
Abstract 222
Introduction 222
Materials and Methods 223
Kohonen Network 224
Template Approach 225
From a 3D-Space to a 2D-Map 226
Clustering of the Structures by an Investigation of their Maps 228
In Search of the Bioactive Conformation, the Best Superposition, SAR 230
Conclusions 233
References 234
Chapter 10. A New Nonlinear Neural Mapping Technique for Visual Exploration of QSAR Data 236
Abstract 236
Introduction 236
Background 238
Case Study I: Analysis of Sensor Data 244
Case Study II: Optimal Test Series Design 250
Concluding Remarks 259
References 259
Chapter 11. Combining Fuzzy Clustering and Neural Networks to Predict Protein Structural Classes 268
Abstract 268
Introduction 269
Methodology 270
Results and Discussion 279
Caveats and Conclusions 288
References 290
Index 294
Color Plate Section 298