Virtual Screening: Principles, Challenges, and Practical Guidelines

List of Contributors XVII

Preface XXIII

A Personal Foreword XXV

Part One Principles 1

1 Virtual Screening of Chemical Space: From Generic Compound Collections to Tailored Screening Libraries 3
Markus Boehm

1.1 Introduction 3

1.2 Concepts of Chemical Space 4

1.3 Concepts of Druglikeness and Leadlikeness 6

1.4 Diversity-Based Libraries 8

1.4.1 Concepts of Molecular Diversity 8

1.4.2 Descriptor-Based Diversity Selection 9

1.4.3 Scaffold-Based Diversity Selection 12

1.4.4 Sources of Diversity 13

1.5 Focused Libraries 15

1.5.1 Concepts of Focused Design 15

1.5.2 Ligand-Based Focused Design 16

1.5.3 Structure-Based Focused Design 17

1.5.4 Chemogenomics Approaches 18

1.6 Virtual Combinatorial Libraries and Fragment Spaces 20

1.7 Databases of Chemical and Biological Information 21

1.8 Conclusions and Outlook 24

1.9 Glossary 25

References 26

2 Preparing and Filtering Compound Databases for Virtual and Experimental Screening 35
Maxwell D. Cummings, Éric Arnoult, Christophe Buyck, Gary Tresadern, Ann M. Vos, and Jörg K. Wegner

2.1 Introduction 35

2.2 Ligand Databases 36

2.2.1 Chemical Data Structures 36

2.2.2 3D Conformations 38

2.2.3 Data Storage 39

2.2.4 Workflow Tools 39

2.2.5 Past Reviews and Recent Papers 40

2.3 Considering Physicochemical Properties 42

2.3.1 Druglikeness 42

2.3.2 Leadlikeness and Beyond 43

2.4 Undesirables 43

2.4.1 Screening Artifacts 44

2.4.2 Pharmacologically Promiscuous Compounds 45

2.5 Property-Based Filtering for Selected Targets 46

2.5.1 Antibacterials 47

2.5.2 CNS 49

2.5.3 Protein–Protein Interactions 51

2.6 Summary 52

References 53

3 Ligand-Based Virtual Screening 61
Herbert Koeppen, Jan Kriegl, Uta Lessel, Christofer S. Tautermann, and Bernd Wellenzohn

3.1 Introduction 61

3.2 Descriptors 62

3.3 Search Databases and Queries 67

3.3.1 Selection of Reference Ligands 67

3.3.2 Preparation of the Search Database 68

3.4 Virtual Screening Techniques 68

3.4.1 Similarity Searches 69

3.4.2 Similarity Searches in Very Large Chemical Spaces 72

3.4.3 Machine Learning in Virtual Screening 74

3.4.4 Validation of Methods and Prediction of Success 78

3.5 Conclusions 79

References 80

4 The Basis for Target-Based Virtual Screening: Protein Structures 87
Jason C. Cole, Oliver Korb, Tjelvar S.G. Olsson, and John Liebeschuetz

4.1 Introduction 87

4.2 Selecting a Protein Structure for Virtual Screening 87

4.2.1 Why Are There Errors in Crystal Structures? 87

4.2.2 Possible Problems That May Occur in a Crystal Structure 91

4.2.3 Structural Relevance 95

4.2.4 Critical Evaluation of Models: Recognizing Issues in Structures 98

4.3 Setting Up a Protein Model for vHTS 101

4.3.1 Binding Site Definition 101

4.3.2 Protonation 104

4.3.3 Treatment of Solvent in Docking 104

4.3.4 Use of Protein-Based Constraints in Docking 105

4.3.5 Protein Flexibility 106

4.4 Summary 109

4.5 Glossary of Crystallographic Terms 110

4.5.1 R-Factor 110

4.5.2 Resolution 110

4.5.3 2mFo-DFc Map 110

References 110

5 Pharmacophore Models for Virtual Screening 115
Patrick Markt, Daniela Schuster, and Thierry Langer

5.1 Introduction 115

5.2 Compilation of Compounds 116

5.2.1 Chemical Structure Generation 116

5.2.2 Conformational Analysis 116

5.3 Pharmacophore Model Generation 117

5.3.1 State of the Art 117

5.3.2 Structure-Based Methods 117

5.3.3 Ligand-Based Methods 118

5.3.4 Limitations of Ligand-Based Methods 119

5.4 Validation of Pharmacophore Models 119

5.4.1 Chemical Databases for Validation 119

5.4.2 Enrichment Assessment 121

5.4.3 Enrichment Metrics 122

5.4.4 Receiver Operating Characteristic Curve Analysis 124

5.4.5 Area Under the ROC Curve 125

5.5 Pharmacophore-Based Screening 127

5.5.1 DS CATALYST 128

5.5.2 UNITY (GALAHAD/GASP) 128

5.5.3 LIGANDSCOUT 129

5.5.4 MOE 130

5.5.5 PHASE 130

5.6 Postprocessing of Pharmacophore-Based Screening Hits 131

5.6.1 Lead- and Druglikeness 131

5.6.2 Structural Similarity Analysis 131

5.7 Pharmacophore-Based Parallel Screening 132

5.8 Application Examples for Synthetic Compound Screening 133

5.8.1 17b-Hydroxysteroid Dehydrogenase 1 Inhibitors 133

5.8.2 Cannabinoid Receptor 2 (CB2) Ligands 134

5.8.3 Further Application Examples 136

5.9 Application Examples for Natural Product Screening 136

5.9.1 Cyclooxygenase (COX) Inhibitors 139

5.9.2 Sigma-1 (s1) Receptor Ligands 139

5.9.3 Acetylcholinesterase Inhibitors 140

5.9.4 Human Rhinovirus Coat Protein Inhibitors 141

5.9.5 Quorum-Sensing Inhibitors 141

5.9.6 Peroxisome Proliferator-Activated Receptor c Ligands 141

5.9.7 b-Ketoacyl-Acyl Carrier Protein Synthase III Inhibitors 142

5.9.8 5-Lipoxygenase Inhibitors 142

5.9.9 11b-Hydroxysteroid Dehydrogenase Type 1 Inhibitors 142

5.9.10 Pharmacophore-Based Parallel Screening of Natural Products 143

5.10 Conclusions 143

References 144

6 Docking Methods for Virtual Screening: Principles and Recent Advances 153
Didier Rognan

6.1 Principles of Molecular Docking 153

6.1.1 Sampling Degrees of Freedom of the Ligand 154

6.1.2 Scoring Ligand Poses 156

6.2 Docking-Based Virtual Screening Flowchart 158

6.2.1 Ligand Setup 158

6.2.2 Protein Setup 159

6.2.3 Docking 160

6.2.4 Postdocking Analysis 161

6.3 Recent Advances in Docking-Based VS Methods 162

6.3.1 Novel Docking Algorithms 162

6.3.2 Fragment Docking 164

6.3.3 Postdocking Refinement 164

6.3.4 Addressing Protein Flexibility 166

6.3.5 Solvated or Dry? 168

6.4 Future Trends in Docking 168

References 169

Part Two Challenges 177

7 The Challenge of Affinity Prediction: Scoring Functions for Structure-Based Virtual Screening 179
Christoph Sotriffer and Hans Matter

7.1 Introduction 179

7.2 Physicochemical Basis of Protein–Ligand Recognition 180

7.3 Classes of Scoring Functions 185

7.3.1 Force Field-Based Methods 185

7.3.2 Empirical Scoring Functions 189

7.3.3 Knowledge-Based Scoring Functions 191

7.4 Interesting New Approaches to Scoring Functions 192

7.4.1 Improved Treatment of Hydrophobicity and Dehydration 192

7.4.2 Development and Validation of SFCscore 194

7.4.3 Consensus Scoring 195

7.4.4 Tailored Scoring Functions 196

7.4.5 Structural Interaction Fingerprints 199

7.5 Comparative Assessment of Scoring Functions 200

7.6 Tailoring Scoring Strategies in Virtual Screening 203

7.6.1 Toward a Strategy for Applying Scoring Functions 203

7.6.2 Retrospective Validation Prior to Prospective Virtual Screening 204

7.6.3 Lessons Learned: Improvements in Scoring Evaluations 205

7.6.4 Postfiltering Results of Virtual Screenings 205

7.7 Caveats for Development of Scoring Functions 206

7.7.1 General Points 206

7.7.2 Biological Data 207

7.7.3 Structural Data on Protein–Ligand Complexes and Decoy Data Sets 207

7.7.4 Cooperativity and Other Model Deficiencies 208

7.8 Conclusions 209

References 210

8 Protein Flexibility in Structure-Based Virtual Screening: From Models to Algorithms 223
Angela M. Henzler and Matthias Rarey

8.1 How Flexible Are Proteins? – A Historical Perspective 223

8.1.1 Ligand Binding Is Coupled with Protein Conformational Change 223

8.1.2 Types of Flexibility 224

8.2 Flexible Protein Handling in Protein–Ligand Docking 225

8.2.1 Docking Following Conformational Selection 227

8.2.2 Induced Fit Docking: Single-Structure-Based Docking Techniques 231

8.2.3 Integrated Docking Approaches 235

8.3 Flexible Protein Handling in Docking-Based Virtual Screening 236

8.3.1 Efficiency of Fully Flexible Docking Approaches in Retrospective 237

8.3.2 Discrimination of Binders and Nonbinders 238

8.4 Summary 238

References 239

9 Handling Protein Flexibility in Docking and High-Throughput Docking: From Algorithms to Applications 245
Claudio N. Cavasotto

9.1 Introduction: Docking and High-Throughput Docking in Drug Discovery 245

9.2 The Challenge of Accounting for Protein Flexibility in Docking 246

9.2.1 Theoretical Understanding of the Problem 246

9.2.2 Docking Failures Due to Protein Flexibility 247

9.3 Accounting for Protein Flexibility in Docking-Based Drug Discovery and Design 250

9.3.1 Receptor Ensemble-Based Docking Methods 252

9.3.2 Single-Structure-Based Docking Methods 253

9.3.3 Multilevel Methods 256

9.3.4 Homology Modeling 257

9.4 Conclusions 257

References 258

10 Consideration of Water and Solvation Effects in Virtual Screening 263
Johannes Kirchmair, Gudrun M. Spitzer, and Klaus R. Liedl

10.1 Introduction 263

10.2 Experimental Approaches for Analyzing Water Molecules 266

10.3 Computational Approaches for Analyzing Water Molecules 271

10.3.1 Molecular Dynamics Simulations 271

10.3.2 Empirical and Implicit Considerations of Solvation Effects 274

10.4 Water-Sensitive Virtual Screening: Approaches and Applications 275

10.4.1 Protein–Ligand Docking 275

10.4.2 Pharmacophore Modeling 278

10.5 Conclusions and Recommendations 281

References 282

Part Three Applications and Practical Guidelines 291

11 Applied Virtual Screening: Strategies, Recommendations, and Caveats 293
Dagmar Stumpfe and Jürgen Bajorath

11.1 Introduction 293

11.2 What Is Virtual Screening? 293

11.3 Spectrum of Virtual Screening Approaches 294

11.4 Molecular Similarity as a Foundation and Caveat of Virtual Screening 295

11.5 Goals of Virtual Screening 296

11.6 Applicability Domain 297

11.7 Reference and Database Compounds 299

11.8 Biological Activity versus Compound Potency 300

11.9 Methodological Complexity and Compound Class Dependence 301

11.10 Search Strategies and Compound Selection 302

11.11 Virtual and High-Throughput Screening 304

11.12 Practical Applications: An Overview 306

11.13 LFA-1 Antagonist 307

11.14 Selectivity Searching 310

11.15 Concluding Remarks 314

References 315

12 Applications and Success Stories in Virtual Screening 319
Hans Matter and Christoph Sotriffer

12.1 Introduction 319

12.2 Practical Considerations 320

12.3 Successful Applications of Virtual Screening 321

12.3.1 Structure-Based Virtual Screening 322

12.3.2 Structure-Based Library Design 336

12.3.3 Ligand-Based Virtual Screening 338

12.4 Conclusion 347

References 348

Part Four Scenarios and Case Studies: Routes to Success 359

13 Scenarios and Case Studies: Examples for Ligand-Based Virtual Screening 361
Trevor Howe, Daniele Bemporad, and Gary Tresadern

13.1 Introduction 361

13.2 1D Ligand-Based Virtual Screening 362

13.3 2D Ligand-Based Virtual Screening 363

13.3.1 Examples from the Literature 363

13.3.2 Applications at J&JPRD Europe 366

13.4 3D Ligand-Based Virtual Screening 368

13.4.1 Methods 370

13.4.2 3DLBVS Examples 372

13.5 Summary 376

References 377

14 Virtual Screening on Homology Models 381
Róbert Kiss and György M. Keseru"

14.1 Introduction 381

14.2 Homology Models versus Crystal Structures: Comparative Evaluation of Screening Performance 382

14.2.1 Soluble Proteins 382

14.2.2 Membrane Proteins 392

14.3 Challenges of Homology Model-Based Virtual Screening 394

14.3.1 Level of Sequence Identity 395

14.3.2 Main-Chain Flexibility 396

14.3.3 Side-Chain Conformation: Induced Fit Effects of Ligands 396

14.3.4 Loop Modeling 397

14.4 Case Studies 399

14.4.1 Virtual Screening on the Homology Model of Histamine H4 Receptor 399

14.4.2 Virtual Screening on the Homology Model of Janus Kinase 2 402

References 404

15 Target-Based Virtual Screening on Small-Molecule Protein Binding Sites 411
Ralf Heinke, Urszula Uciechowska, Manfred Jung, and Wolfgang Sippl

15.1 Introduction 411

15.1.1 Pharmacophore-Based Methods 412

15.1.2 Ligand Docking 412

15.1.3 Virtual Screening 413

15.1.4 Binding Free Energy Calculations 414

15.2 Structure-Based VS for Histone Arginine Methyltransferase PRMT1 Inhibitors 414

15.2.1 Structure-Based VS of the NCI Diversity Set 415

15.2.2 Pharmacophore-Based VS 417

15.3 Identification of Nanomolar Histamine H3 Receptor Antagonists by Structure- and Pharmacophore-Based VS 422

15.3.1 Generation of Homology Model of the hH3R and hH3R Antagonist Complexes 423

15.3.2 Validation of the Homology Model by Docking Known Antagonists into the hH3R Binding Site 424

15.3.3 Pharmacophore-Based VS 425

15.3.4 Experimental Testing of the Identified Hits 429

15.3.5 Discussion of the Applied VS Strategies 429

15.4 Summary 431

References 432

16 Target-Based Virtual Screening to Address Protein–Protein Interfaces 435
Olivier Sperandio, Maria A. Miteva, and Bruno O. Villoutreix

16.1 Introduction 435

16.2 Some Recent PPIM Success Stories 437

16.3 Protein–Protein Interfaces 438

16.3.1 Interface Pockets, Flexibility, and Hot Spots 440

16.3.2 Databases and Tools to Analyze Interfaces 442

16.4 PPIMs. Chemical Space and ADME/Tox Properties 442

16.5 Drug Discovery, Chemical Biology, and In Silico Screening Methods: Overview and Suggestions for PPIM Search 447

16.6 Case Studies 450

16.6.1 PPI Stabilizers: Superoxide Dismutase Type 1 450

16.6.2 PPI Inhibitors: Lck 452

16.6.3 Allosteric Inhibitors: Antitrypsin Polymerization 455

16.7 Conclusions and Future Directions 457

References 458

17 Fragment-Based Approaches in Virtual Screening 467
Danzhi Huang and Amedeo Caflisch

17.1 Introduction 467

17.2 In Silico Fragment-Based Approaches 468

17.3 Our Approach to High-Throughput Fragment-Based Docking 470

17.3.1 Decomposition of Compounds into Fragments 471

17.3.2 Docking of Anchor Fragments 471

17.3.3 Flexible Docking of Library Compounds 472

17.3.4 LIECE Binding Energy Evaluation 472

17.3.5 Consensus Scoring 475

17.3.6 In Silico Screening Campaigns 475

17.3.7 West Nile Virus NS3 Protease (Flaviviral Infections) 475

17.3.8 EphB4 Tyrosine Kinase (Cancer) 477

17.4 Lessons Learned from Our Fragment-Based Docking 479

17.5 Challenges of Fragment-Based Approaches 481

References 482

Appendix A: Software Overview 491

Appendix B: Virtual Screening Application Studies 501

Index 511