Chemoinformatics approaches to virtual screening

摘要: Preface 1 - Fragment Descriptors in SAR/QSAR/QSPR studies, molecular similarity analysis and virtual screening Introduction Historical survey Main characteristics of Types Fragments Simple Fixed WLN SMILES Atom-Centered Bond-Centered Maximum Common Substructures Atom Pairs Topological Multiplets Substituents Molecular Frameworks Basic Subgraphs Mined Random Library describing supramolecular systems chemical reactions Storage fragments' information Fragment's Connectivity Generic Graphs Labeling Atoms Application Virtual Screening In Silico Design Filtering Similarity Search SAR Classification (Probabilistic) Models QSAR/QSPR Regression Limitations Conclusion 2 Pharmacophores 3D pharmacophore models descriptors pharmacophores from 2D-aligments 2D fingerprints index-based 'pharmacophores'? pair triplets searching with Technical Issues Some Examples Machine-learning Fingerprints Conclusions 3 Pharmacophore-based Drug Discovery Methods Chemical Feature-based The Term "3D Pharmacophore" Feature Definitions Pharmacophore Representation Hydrogen bonding interactions Lipophilic areas Aromatic Charge-transfer Customization definition new features Current super-positioning techniques for aligning molecules Generation Use Ligand-based Modeling Structure-based Inclusion Shape Information Qualitative vs. Quantitative Validation as Part a Multi-Step Approach Antitarget ADME(T) Using Activity Profiling Parallel Method Extensions Comparisons to Other Shape-based Docking Constraints Used Further Reading Summary 4 Analysis Ligand-Based Foundations Spaces Measures Landscapes Analyzing the Nature Structure-Activity Relationships between different SARs target-ligand characterization Implications Strengths Future Perspectives 5 Field Topology drug design Introduction: local parameters QSAR, Supergraph-based QSAR Rationale history (MFTA) General principles Local descriptors: facets ligand-biotarget interaction Construction supergraph Formation descriptor matrix Statistical Applicability control From MFTA model biotarget action MFTA-based compound databases generated structure libraries 6 Probabilistic approaches activity prediction Biological Dose-Effect Experimental Data Preparation Training Sets Creation Evaluation Mathematical Approaches Prediction Accuracy Single-Targeted Multi-Targeted PASS Activities Predicted by Structure Description Base Algorithm Spectrum Estimation Interpretation Results Selection Most Prospective Compounds 7 Fragment-based de novo druglike Molecules Scoring Outlook 8 Early ADME/T predictions: toy or tool? Which properties are important early discovery? Physico-chemical profiling Lipophilicity Solubility availability accuracy Why don't work: challenge Domain AD based on space property-based How reliable predictions physico-chemical properties? Available biological Absorption Distribution usefulness is limited available data 9 Compound Principles Applications Specific Guided Targeted Libraries Diversity Based Focused Protein Developability Drug-likeness Rule & Alert ADMET Undesirable Functionality Filters Multiple Objectives Targets Simultaneously Concluding Remarks 10 Integrated Chemo- Bioinformatics Availability large collections NIH Roadmap Initiative PubChem database public domain Major methodologies Challenges limitations current implementation cheminformatics concepts Predictive tools Critical Importance validation domains acceptability criteria modeling workflow application protein ligand interface: EnTESS method Derivation binding affinity approach screening: CoLiBRI representation three-dimensional active sites multidimensional chemistry mapping spaces ligands