- Neu
E-Book, Englisch, 248 Seiten
Reihe: Bentham Science Publishers
Grewal / Lather / Deswal Computer-Aided Drug Design: Principles, Techniques, and Applications: Part 2
1. Auflage 2026
ISBN: 979-8-89881-825-8
Verlag: Bentham Science Publishers
Format: EPUB
Kopierschutz: 0 - No protection
E-Book, Englisch, 248 Seiten
Reihe: Bentham Science Publishers
ISBN: 979-8-89881-825-8
Verlag: Bentham Science Publishers
Format: EPUB
Kopierschutz: 0 - No protection
Computer-aided drug design (CADD) has rapidly expanded far beyond its traditional applications through greater computing power, sophisticated algorithms, large molecular databases and artificial intelligence. This volume presents advanced methodologies shaping modern pharmaceutical research, including de novo design, virtual screening, pharmacophore mapping, drug repurposing, machine learning and molecular dynamics simulations. Tutorials and software-based exercises help readers apply concepts directly, while expert contributors from pharmaceutical sciences, computational chemistry and bioinformatics connect advanced theory with real-world drug discovery practice.
Key Features
Covers de novo drug design, virtual screening and pharmacophore mapping
Explains computational drug repurposing strategies
Applies deep learning and reinforcement learning to CADD
Details molecular dynamics for drug-target interactions
Includes tutorials and software-based exercises
Features expert, multidisciplinary contributors
Readership
Researchers, educators and graduate students in pharmaceutical sciences, medicinal chemistry, computational chemistry and bioinformatics, plus drug discovery professionals.
Autoren/Hrsg.
Weitere Infos & Material
Designing Drugs from Scratch: The Computational and AI Revolution in De Novo Drug Discovery
Shaikh Mohmed Adnan Mohmed Javid1, Sonam M. Gandhi1, Nemat Ali2, Devesh U. Kapoor3, *
Abstract
The computational design of new drugs (also known as de novo drug design) enables researchers to create a completely new therapeutic molecule without requiring any previous knowledge of an existing drug or compound for inspiration, through the use of Advanced Algorithms and Structural Information. The methodologies and techniques for this type of molecular discovery are thoroughly covered in this chapter and introduce readers to the foundational Principles and Importance of Drug Discovery. The methodology section details the various approaches (both structure-based and ligand-based) that can be taken when performing de novo drug design, as well as the atom- and fragment-based methods for constructing molecules. This chapter details the importance of Sampling Methods in the exploration of Chemical Space and the need for various sampling methods (i.e., Monte Carlo, Genetic Algorithms, etc.) in order to produce a diverse library of potential compounds. As such, this chapter will serve as a useful reference for both New and Experienced Researchers and/ or Professionals interested in learning about cutting-edge methodologies of drug discovery via de novo design and their application to the Research and Development of Therapeutic Agents (i.e., Drugs). The latter section discusses the use of Evolutionary Algorithms and related concepts based on Evolutionary Principles (i.e., adaptation based on natural selection) that are beneficial for optimising Molecular Design. In addition, this chapter will provide an overview of the emergent technologies of Deep Reinforcement Learning (DRL) and will discuss the potential of these technologies to transform the Automation and Optimization of Drug Discovery Processes. Finally, in the Future Perspectives section, we will discuss Emerging Trends, Challenges, and Opportunities related to the use of DRL for Drug Design.
* Corresponding author Devesh U. Kapoor: Dr. Dayaram Patel Pharmacy College, Bardoli-394601, Gujarat, India; E-mail: dev7200@gmail.com
INTRODUCTION
De novo drug design is a sophisticated and intricate process that involves the generation of entirely new lead compounds characterized by desirable pharmacological effects and optimal physicochemical properties conducive to therapeutic applications [1, 2]. These advanced computational models possess the remarkable ability to synthesize and generate a wide array of molecular structures that exhibit particular and predetermined physical or biological characteristics, achieved through the sophisticated integration of both predictive and generative neural network architectures that work in tandem to optimize the molecular design process [3, 4].
PRINCIPLES OF DE NOVO DRUG DESIGN
The fundamental concept of de novo drug design pertains to the synthesis of innovative drug-like entities from a foundational level, employing sophisticated computational methodologies and artificial intelligence. This methodology aspires to enhance the drug discovery paradigm by producing compounds endowed with targeted pharmacological characteristics, consequently diminishing the temporal and material investments customarily necessitated by conventional approaches. The subsequent sections delineate critical components of de novo drug design [5, 6]. The approach to designing drugs based on the structure begins with an in-depth grasp of the spatial arrangement of a protein of interest. This particular protein acts as the cornerstone for the identification of potential drug-binding sites [7].
De Novo Software and Its Methodology
De novo software tools are essential in various scientific disciplines for creating novel molecules, genetic assemblies, and peptide sequences without relying on existing templates. These tools employ various methodologies, from rule-based systems to modern artificial-intelligence systems. In drug discovery, AutoGrow4 employs a genetic algorithm to evolve small molecules through mutation and crossover, optimizing based on docking scores [8]. REINVENT integrates reinforcement learning with SMILES-based generative models to produce molecules with desired pharmacological profiles [9]. Another approach, the deep graph molecular generator, utilizes graph neural networks to simultaneously generate compounds and predict their properties [10]. The drug AI platform merges transformer-based deep learning with the Monte Carlo Tree Search to efficiently explore the chemical space [11]. In genomics, SOAPdenovo-Trans assembles transcriptomes from RNA-Seq data without a reference genome using a de Bruijn graph model to handle alternative splicing and gene variation [12]. HipMer scales this task for large genomes by using parallel processing to assemble the sequences on supercomputers [13]. For metagenomics, DIME applies a divide-and-conquer strategy by clustering reads and merging partial assemblies to improve accuracy and performance. In proteomics, Point Novo leverages deep neural networks to predict peptide sequences directly from mass spectra, thereby enhancing performance across different resolutions [14]. Contra Novo applies contrastive learning to improve the identification of peptide-spectrum matches [15], whereas Vonode uses mass accuracy scoring to improve de novo peptide identification [16]. These tools highlight the growing integration of AI and high-performance computing in biological design and analysis, thereby driving innovation in drug development and genomics and proteomics research.
Determining the Protein Structure
The preliminary phase of the drug discovery continuum entails the acquisition of structural information on the target protein. Using techniques such as X-ray crystallography, Nuclear Magnetic Resonance (NMR), or cryo-electron microscopy [17], researchers can uncover the shape and surface details of a protein, helping them understand how it functions and interacts with other molecules. The identification of pivotal pockets and grooves within proteins is essential for the formulation of molecules capable of effective interactions [18].
Modeling the Binding Site
When the protein structure is available, computational models are used to delineate and scrutinize the binding sites [7]. The procedure includes thorough mapping of the chemical and physical features of the site, incorporating hydrophobic segments, hydrogen bond donors and acceptors, and electrostatic interactions [19]. By constructing a model of this pocket, investigators can forecast which categories of molecules may fit and engage optimally [7].
Designing a Molecule: That Fits. Grounded in the binding site model, chemists and computational biologists collaborate to engineer prospective drug candidates. These compounds are meticulously designed to align with the shape and chemical characteristics of the pocket, thereby ensuring a robust and specific interaction [7]. Utilizing computational docking and molecular dynamics simulations, the stability and binding efficacy of these compounds are assessed prior to progressing to laboratory experimentation [19].
Defining the Concept of De Novo Drug Design
De novo drug design is a sophisticated computational methodology that enables the identification of innovative therapeutic agents that are independent of pre-existing molecular frameworks. This methodology encompasses the systematic design of drug candidates based on the molecular architecture of a target, which is commonly a protein or enzyme implicated in a particular pathological condition. In contrast to conventional drug discovery techniques that modify established compounds,...




