Preface.

1 Molecular Objects and Design Objectives.

1.1 What is a Molecule?

1.2 Simplistic Molecular Representations.

1.3 The Molecular Surface.

1.4 Molecular Shape.

1.5 The Topological Molecular Graph.

1.6 Molecular Properties and Graph Invariants.

1.7 The Drug-likeness Concept.

1.8 Scaffolds, Linkers, and Side-chains.

1.9 Substructure Similarity and "Privileged Motifs".

1.10 Molecules as Strings.

1.11 Constructing Molecules from Strings.

1.12 From Elements to Atom Types.

1.13 Entering the Third Dimension: Automatic Conformer Generation.

1.14 The "Bioactive" Conformation.

Literature.

2 Receptor–Ligand Interaction.

2.1 The Thermodynamics of Protein–Ligand Interaction.

2.2 The Entropic Contribution.

2.3 From Theory to Experiment: Ki and IC50.

2.4 QSAR: Estimating Quantitative Structure–Activity Relationships.

2.4.1 Free–Wilson Analysis.

2.4.2 The Hansch Model.

2.4.3 3D-QSAR.

2.5 Types of Receptor–Ligand Interaction.

2.6 The "Biophore" Concept.

2.7 Potential Pharmacophoric Points.

2.8 The Correlation Vector Approach to Pharmacophore Modeling.

2.9 "Hard Sphere" and "Fuzzy" Pharmacophore Models.

2.10 Lessons from Automated Ligand Docking and Scoring: What Works and What Does Not.

2.11 Fits Like a Glove: Alternative Ligand Binding Modes and Induced Fit Effects.

Literature.

3 Creating the Design.

3.1 Why We Need Computer-assisted Molecular Design.

3.2 The Number of Drug Targets is Limited.

3.3 Ligand Binding Sites.

3.4 Ligand-based Design of Compound Libraries.

3.5 Similar Compounds Do Not Necessarily Interact with Their Target in Similar Ways.

3.6 The Same Ligand Can Adopt Multiple Binding Modes.

3.7 GPCRs Represent a Challenging Target Family.

3.8 Natural Products Are a Source of Inspiration.

3.9 Transition State Analogs Are Potent Enzyme Inhibitors.

3.10 New Targets Sometimes Require a New Ligand Design Concept.

3.11 De novo Design Concepts.

3.12 Primary and Secondary Constraints in de novo Design.

Literature.

4 Virtual Screening Triage.

4.1 The Drug Discovery Pipeline.

4.2 High-throughput Screening (HTS): Why Is It Successful?

4.3 From Hit to Lead.

4.4 Rationalizing the Design Process.

4.5 From High to Low Diversity.

4.6 Quantifying Diversity is Difficult.

4.7 From Negative Design to Positive Design.

4.8 Watch Out for Frequent Hitters!

4.9 Shape-matching: A Coarse-grained Filtering Step.

4.10 The Ultimate Goal: Scaffold-hopping.

4.11 Assessing Chemotype Diversity in Focused Libraries.

4.12 It Works! Examples of Successful Scaffold-hops Found by Virtual Screening.

4.13 Case Studies.

4.13.1 Design of Kv1.5 Ion Channel Modulators.

4.13.2 Virtual Screening of a Natural-product-derived Combinatorial Library for Novel 5-Lipoxygenase Inhibitors.

4.13.3 Scaffold de novo Design for Cannabinoid-1 (CB-1) Receptor Ligands.

Literature.

5 Secondary Design Constraints and Machine Learning.

5.1 Physicochemistry and Pharmacokinetics.

5.2 The "Rule of 5".

5.3 Pharmacokinetics.

5.4 Absorption.

5.5 Distribution.

5.6 Metabolism.

5.7 Elimination.

5.8 Toxicity.

5.9 Prodrugs and Bioisosteres.

5.10 Machine Learning Methods Support Lead Finding and Optimization.

5.11 An Important Step: Data Scaling.

5.12 Application of Machine Learning to Compound Library Design.

5.13 A "Pharmacophore Road Map".

5.14 Case Studies.

5.14.1 Predicting Cross-activities of Allosteric Modulators of Metabotropic Glutamate Receptors (mGluR).

5.14.2 Dopamine D3 Antagonists and ACE Inhibitors.

5.14.3 An Artificial Ant System for Combinatorial Optimization.

Literature.

Subject Index.