Drug Design Using Machine Learning

Cover
Half-Title Page
Series Page
Title Page
Copyright Page
Contents
Preface
1 Molecular Recognition and Machine Learning to Predict Protein-Ligand Interactions
	1.1 Introduction
		1.1.1 Molecular Recognition
	1.2 Molecular Docking
		1.2.1 Conformational Search Algorithm
		1.2.2 Scoring Function with Conventional Methods
	1.3 Machine Learning
		1.3.1 Machine Learning in Molecular Docking
		1.3.2 Machine Learning Challenges in Molecular Docking
	1.4 Conclusions
	References
2 Machine Learning Approaches to Improve Prediction of Target-Drug Interactions
	2.1 Machine Learning Revolutionizing Drug Discovery
		2.1.1 Introduction
		2.1.2 Virtual Screening and Rational Drug Design
		2.1.3 Small Organic Molecules and Peptides as Drugs
	2.2 A Brief Summary of Machine Learning Models
		2.2.1 Support Vector Machines (SVM)
		2.2.2 Random Forests (RF)
		2.2.3 Gradient Boosting Decision Tree
		2.2.4 K-Nearest Neighbor (KNN)
		2.2.5 Neural Network and Deep Learning
		2.2.6 Gaussian Process Regression
		2.2.7 Evaluating Regression Methods
		2.2.8 Evaluating Classification Methods
	2.3 Target Validation
		2.3.1 Ligand Binding Site Prediction (LBS)
		2.3.2 Classical Approaches
		2.3.3 Machine Learning Approaches
			2.3.3.1 SVM-Based Approaches
			2.3.3.2 Random Forest–Based Approaches
			2.3.3.3 Deep Learning–Based Approaches
	2.4 Lead Discovery
		2.4.1 The Relevance of Predict Binding Affinity
		2.4.2 The Concept of Docking
		2.4.3 The Scoring Function
		2.4.4 Developing of Novels Scoring Functions by Machine Learning
			2.4.4.1 Random Forests
			2.4.4.2 Support Vector Machines
			2.4.4.3 Neural Networks
			2.4.4.4 Gradient Boosting Decision Tree
	2.5 Lead Optimization
		2.5.1 QSAR and Proteochemometrics
		2.5.2 Machine Learning Algorithms in Deriving Descriptors
	2.6 Peptides in Pharmaceuticals
		2.6.1 Peptide Natural and Synthetic Sources
		2.6.2 Applications and Market for Peptides-Based Drugs
		2.6.3 Challenges to Become a Peptide Into a Drug
		2.6.4 Improving Peptide Drug Development Using Machine Learning Techniques
	2.7 Conclusions
	References
3 Machine Learning Applications in Rational Drug Discovery
	3.1 Introduction
	3.2 The Drug Development and Approval Process
	3.3 Human-AI Partnership
	3.4 AI in Understanding the Pathway to Assess the Side Effects
		3.4.1 Traditional Versus New Strategies in Drug Discovery
		3.4.2 Target Identification and Authentication
		3.4.3 Searching the Hit and Lead Molecules with the Help of AI
		3.4.4 Discretion of a Population for Medical Trials Using AI
	3.5 Predicting the Side Effects Using AI
	3.6 AI for Polypharmacology and Repurposing
	3.7 The Challenge of Keeping Drugs Safe
	3.8 Conclusion
	Resources
	References
4 Deep Learning for the Selection of Multiple Analogs
	4.1 Introduction
	4.2 Goals of Analog Design
	4.3 Deep Learning in Drug Discovery
	4.4 Chloroquine Analogs
	4.5 Deep Learning in Medical Field
		4.5.1 Scientific Study of Skin Diseases
		4.5.2 Anatomical Laparoscopy
		4.5.3 Angiography
		4.5.4 Interpretation of Wound
		4.5.5 Molecular Docking
		4.5.6 Breast Cancer Detection
		4.5.7 Polycystic Organs
		4.5.8 Bone Tissue
		4.5.9 Interaction Drug-Target
		4.5.10 Pancreatic Issue Prediction
		4.5.11 Prediction of Carcinoma in Cells
		4.5.12 Determining Parkinson’s
		4.5.13 Segregating Cells
	4.6 Conclusion
	References
5 Drug Repurposing Based on Machine Learning
	5.1 Introduction
	5.2 Computational Drug Repositioning Strategies
		5.2.1 Drug-Based Strategies
		5.2.2 Disease-Based Strategies
	5.3 Machine Learning
	5.4 Data Resources Used for Computational Drug Repositioning Through Machine Learning Techniques
	5.5 Machine Learning Approaches Used for Drug Repurposing
		5.5.1 Network-Based Approaches
		5.5.2 Text Mining-Based Approaches
		5.5.3 Semantics-Based Approaches
	5.6 Drugs Repurposing Through Machine Learning-Case Studies
		5.6.1 Psychiatric Disorders
		5.6.2 Alzheimer’s Disease
		5.6.3 Drug Repurposing for Cancer
		5.6.4 COVID-19
		5.6.5 Herbal Drugs
	5.7 Conclusion
	References
6 Recent Advances in Drug Design With Machine Learning
	6.1 Introduction
	6.2 Categorization of Machine Learning Tasks
		6.2.1 Supervised Learning
		6.2.2 Unsupervised Learning
		6.2.3 Semisupervised Learning
		6.2.4 Reinforcement Learning
	6.3 Machine Language-Mediated Predictive Models in Drug Design
		6.3.1 Quantitative Structure-Activity Relationship Models (QSAR)
		6.3.2 Quantitative Structure-Property Relationship Models (QSPR)
		6.3.3 Quantitative Structure Toxicity Relationship Models (QSTR)
		6.3.4 Quantitative Structure Biodegradability Relationship Models (QSBR)
	6.4 Machine Learning Models
		6.4.1 Artificial Neural Networks (ANNs)
		6.4.2 Self-Organizing Map (SOM)
		6.4.3 Multilayer Perceptrons (MLPs)
		6.4.4 Counter Propagation Neural Networks (CPNN)
		6.4.5 Bayesian Neural Networks (BNNs)
		6.4.6 Support Vector Machines (SVMs)
		6.4.7 Naive Bayesian Classifier
		6.4.8 K Nearest Neighbors (KNN)
		6.4.9 Ensemble Methods
			6.4.9.1 Boosting
			6.4.9.2 Bagging
		6.4.10 Random Forest
		6.4.11 Deep Learning
		6.4.12 Synthetic Minority Oversampling Technique
	6.5 Machine Learning and Docking
		6.5.1 Scoring Power
		6.5.2 Ranking Power
		6.5.3 Docking Power
		6.5.4 Predicting Docking Score Using Machine Learning
	6.6 Machine Learning in Chemoinformatics
	6.7 Challenges and Limitations for Machine Learning in Drug Discovery
	6.8 Conclusion and Future Perspectives
	References
7 Loading of Drugs in Biodegradable Polymers Using Supercritical Fluid Technology
	7.1 Introduction
	7.2 Supercritical Fluid Technology
		7.2.1 Supercritical Fluids
		7.2.2 Physicochemical Properties
		7.2.3 Carbon Dioxide
	7.3 Biodegradable Polymers
		7.3.1 Main Biologically-Derived Polymers Used With SCF Technologies
			7.3.1.1 Cellulose
			7.3.1.2 Chitosan
			7.3.1.3 Alginate
			7.3.1.4 Collagen
		7.3.2 Main Synthetic Polymers Used With SCF Technologies
			7.3.2.1 Polylactic Acid (PLA)
			7.3.2.2 Poly (Lactic-co-Glycolic Acid) (PLGA)
			7.3.2.3 Polycaprolactone (PCL)
			7.3.2.4 Poly (Vinyl Alcohol) (PVA)
	7.4 Drug Delivery
		7.4.1 Types of Drugs
		7.4.2 Influence of Experimental Conditions on the Drug Loading
	7.5 Conclusion
	Acknowledgments
	References
8 Neural Network for Screening Active Sites on Proteins
	8.1 Introduction
	8.2 Structural Proteomics
		8.2.1 PPIs
		8.2.2 Active Sites in Proteins
	8.3 Gist Techniques to Study the Active Sites on Proteins
		8.3.1 In Vitro
			8.3.1.1 Affinity Purification
			8.3.1.2 Affinity Chromatography
			8.3.1.3 Coimmunoprecipitation
			8.3.1.4 Protein Arrays
			8.3.1.5 Protein Fragment Complementation
			8.3.1.6 Phage Display
			8.3.1.7 X-Ray Crystallography
			8.3.1.8 Nuclear Magnetic Resonance Spectroscopy (NMR)
		8.3.2 In Vivo
			8.3.2.1 In-Silico Two-Hybrid
		8.3.3 In-Silico and Neural Network
			8.3.3.1 Data Base
			8.3.3.2 Sequence-Based Approaches
			8.3.3.3 Structure-Based Approaches
			8.3.3.4 Phylogenetic Tree
			8.3.3.5 Gene Fusion
	8.4 Neural Networking Algorithms to Study Active Sites on Proteins
		8.4.1 PDBSiteScan Program
		8.4.2 Patterns in Nonhomologous Tertiary Structures (PINTS)
		8.4.3 Genetic Active Site Search (GASS)
		8.4.4 Site Map
		8.4.5 Computed Atlas of Surface Topography of Proteins (CASTp)
	8.5 Conclusion
	References
9 Protein Redesign and Engineering Using Machine Learning
	9.1 Introduction
	9.2 Designing Sequence-Function Model Through Machine Learning
		9.2.1 Training of Model and Evaluation
		9.2.2 Representation of Proteins by Vector
		9.2.3 Guiding Exploration by Employing Sequence-Function Prediction
	9.3 Features Based on Energy
	9.4 Features Based on Structure
	9.5 Prediction of Thermostability of Protein with Single Point Mutations
	9.6 Selection of Features
		9.6.1 Extraction of Features
	9.7 Force Field and Score Function
	9.8 Machine Learning for Prediction of Hot Spots
		9.8.1 Support Vector Machines
		9.8.2 Nearest Neighbor
		9.8.3 Decision Trees
		9.8.4 Neural Networks
		9.8.5 Bayesian Networks
		9.8.6 Ensemble Learning
	9.9 Deep Learning—Neural Network in Computational Protein Designing
	9.10 Machine Learning in Engineering of Proteins
	9.11 Conclusion
	References
10 Role of Transcriptomics and Artificial Intelligence Approaches for the Selection of Bioactive Compounds
	10.1 Introduction
	10.2 Types of Bioactive Compounds
		10.2.1 Phenolic Acids
		10.2.2 Stilbenes
		10.2.3 Ellagitannins
		10.2.4 Flavonoids
		10.2.5 Proanthocyanidin
		10.2.6 Vitamins
		10.2.7 Bioactive Peptides
	10.3 Transcriptomics Approaches for the Selection of Bioactive Compounds
		10.3.1 Hybrid Transcriptome Sequencing
		10.3.2 Microarray
		10.3.3 RNA-Seq
	10.4 Artificial Intelligence Approaches for the Selection of Bioactive Compounds
		10.4.1 Machines Learning (ML) Approach for the Selection of Bioactive Compounds
			10.4.1.1 Evolution of Machine Learning to Deep Learning
			10.4.1.2 Virtual Screening
			10.4.1.3 Recent Advances in Machine Learning
			10.4.1.4 Deep Learning
		10.4.2 De Novo Synthesis of Bioactive Compounds
			10.4.2.1 Application Examples of De Novo Design
		10.4.3 Applications of Machine Learning and Deep Learning
			10.4.3.1 Application of Deep Learning in Compound Activity and Property Prediction
			10.4.3.2 Application of Deep Learning in Biological Imaging Analysis
			10.4.3.3 Future Development of Deep Learning in Drug Discovery
	10.5 Applications of Transcriptomic and Artificial Intelligence Techniques for Drug Discovery
	10.6 Conclusion and Perspectives
	References
11 Prediction of Drug Toxicity Through Machine Learning
	11.1 Introduction
	11.2 Drug Discovery
		11.2.1 Target Identification
		11.2.2 Lead Discovery: Preclinical
		11.2.3 Medicinal Chemistry: Preclinical
		11.2.4 In Vitro Studies
		11.2.5 In Vivo Studies
		11.2.6 Clinical Trials
		11.2.7 Food and Drug Administration Approval
	11.3 Drug Design Through New Techniques
	11.4 Machine Learning as a Science
		11.4.1 Supervised Machine Learning
		11.4.2 Unsupervised Machine Learning
	11.5 Reinforcement Machine Learning
	11.6 AI Application in Drug Design
	11.7 Machine Learning Methods Used in Drug Discovery
		11.7.1 Support Vector Machines
		11.7.2 Random Forest
		11.7.3 Multilayer Perception (MLP)
	11.8 Deep Learning (DL)
	11.9 Drug Design Applications
	11.10 Drug Discovery Problems
		11.10.1 Prognostic Biomarkers
		11.10.2 Digital Pathology
	11.11 Conclusion
	References
12 Artificial Intelligence for Assessing Side Effects
	12.1 Introduction
	12.2 Background
	12.3 Traditional Approach to Pharmacovigilance and Its Limitations
	12.4 Role of Artificial Intelligence in Pharmacological Profiling for Safety Assessment
	12.5 Artificial Intelligence for Assessing Side Effects
	12.6 Conclusion
	References
Index