Skip to content

History

DRAFT

This page is a work in progress and is subject to change at any moment.

Inception of Computational Techniques

  • The application of quantum mechanics to theoretical chemistry in the late 1950s and 1960s, as discussed by Mulliken and Roothaan (1959), highlighted the potential of quantum mechanics for understanding molecular structures and interactions. This marked a critical advancement in the computational approach to drug discovery (Mulliken & Roothaan, 1959).

"The calculation of small molecular interactions by the differences of separate total energies. Some procedures with reduced errors" by S. F. Boys and F. Bernardi (1970): This paper presents a new method for computing molecular interactions, focusing on reducing errors in interaction energy calculations. This advancement was crucial for accurately predicting molecular behaviors and interactions in various chemical and biological processes Boys & Bernardi, 1970.

"Equation of state calculations by fast computing machines" by N. Metropolis, A. W. Rosenbluth, M. Rosenbluth, A. H. Teller (1953): This early work described a general method for investigating the properties of substances consisting of interacting molecules, using a modified Monte Carlo integration over configuration space. The method was suitable for fast computing machines of the time, marking a significant step towards modern computational chemistry Metropolis et al., 1953.

"Studies in Molecular Dynamics. I. General Method" by B. Alder and T. Wainwright (1959): This paper outlines a method for calculating the behavior of several hundred interacting classical particles, marking an early step towards the field of molecular dynamics. This technique allowed for the exact computational study of many-body problems, contributing significantly to the understanding of molecular interactions Alder & Wainwright, 1959.

"Liquid Structure and Self‐Diffusion" (1966) by A. Rahman: This paper used the method of molecular dynamics to study the structure of a liquid and its relation to the process of self-diffusion, introducing a precise geometrical procedure to separate particles around a given particle into shells of primary, secondary, etc., neighbors (Rahman, 1966).

"Molecular Dynamics Studies of the Microscopic Properties of Dense Fluids" (1969) by P. L. Fehder: This paper reported molecular dynamics calculations on a two-dimensional system of Lennard-Jones disks, providing equilibrium thermodynamics data for various temperature-density states and examining the presence of large vacancies in the spatial distribution of particles in the system (Fehder, 1969).

  • The early 1970s saw the development of versatile and interactive graphics display systems for molecular modeling, as exemplified by the work of Feldman et al. (1973). These systems facilitated the study of macromolecules, marking a significant step forward in the field of CADD (Feldman et al., 1973).
  • The late 1970s introduced systems like AIMS (Ames Interactive Molecular modeling System), which utilized 3-D dynamic computer display systems for constructing molecular models and were instrumental in the study of molecular interactions and conformational changes (Coeckelenbergh et al., 1978).

Rise of Structure-Based Drug Design

  • Biophysical Applications of MD: Berendsen's work emphasized the application of MD simulations to complex molecular systems, highlighting its capabilities in predicting the free energy of binding between inhibitors and enzymes. This was instrumental in the application of simulation methods in drug design, demonstrating the potential of MD to contribute significantly to the field (Berendsen, 1987).

  • Chemical and Biomolecular Systems Simulations: Beveridge and Jorgensen's work addressed the simulation of chemical and biomolecular systems, encompassing areas such as free energy simulations and the study of DNA reactions. Their contributions provided a comprehensive overview of the application of computer simulations in understanding biomolecular systems (Beveridge & Jorgensen, 1987).

  • Molecular Dynamics of Proteins: Karplus and colleagues' pioneering efforts in simulating the dynamics of proteins laid the groundwork for the application of MD in studying biomolecules. Their simulations of the bovine pancreatic trypsin inhibitor marked the beginning of molecular dynamics studies of biological macromolecules, setting the stage for future research in protein dynamics and function (Karplus et al., 1987).

  • Constant Pressure and Temperature Simulations: Andersen's development of methods to perform MD simulations under conditions of constant temperature and/or pressure expanded the scope of MD simulations. This allowed for more realistic simulations of biological processes, facilitating the study of biomolecules under varied environmental conditions (Andersen, 1980).

  • Computer-aided Molecular Design: The application of MD simulations in computer-aided molecular design was significantly advanced by Richards, who demonstrated the potential of combining computer graphics techniques with theoretical calculations. This approach was crucial for suggesting molecules with desired specific properties, aiding in the synthetic and chemical manipulation of therapeutic drugs (Richards, 1985).

  • Further developments in the 1970s, including more accurate molecular dynamics simulations, exemplified by Stillinger and Rahman's work on liquid water, underscored the increasing sophistication of computational models in studying molecular properties and interactions (Stillinger & Rahman, 1974).

  • Advances in X-ray Crystallography: The use of high-flux X-ray and neutron solution scattering became instrumental for structural studies of proteins, offering a complement to crystallographic investigations with low-resolution structural methods. This period marked an increase in the quantitative measurements of macromolecular structures and dynamics (Perkins, 1988).

  • Biomolecular Dynamics and Crystallography: Workshops and collaborative efforts underscored the emerging picture of biomolecular dynamics, supported by crystallographic data. This led to a better understanding of enzyme catalysis, nucleic acid functions, and membrane transport, revealing the time dimension in biomolecular interactions (Edholm et al., 1984).

  • Refinement Techniques and Structural Determination: The development of molecular dynamics for the refinement of macromolecular structures demonstrated the feasibility of achieving high-resolution data, critical for understanding biotin-avidin interactions and other complex biomolecular mechanisms (Hendrickson et al., 1989).

  • Emergence of Biological Crystallogenesis: The concept of purity and methodological principles in biological crystallogenesis gained attention, emphasizing the need for a more rational approach to the crystallization of biomacromolecules and their complexes. This was pivotal for the growth of crystals for structural analysis (Giegé & Mikol, 1989).

  • The Protein Data Bank (PDB): The establishment of the PDB as a computer-based archival file for macromolecular structures marked a significant milestone, facilitating the storage and public distribution of atomic coordinates and structural data for a wide array of biomolecules (Bernstein et al., 1977).

  • Fast Energy Estimation and Visualization of Protein-Ligand Interaction: A new computational and graphical method was developed to aid ligand-protein docking studies, capable of estimating non-bonded and electrostatic interaction energy in real-time during interactive docking operations. This method also allowed for the visualization of the local environment inside the binding pocket, significantly aiding the drug design process (Tomioka, Itai, & Iitaka, 1987).

  • Docking Software Development: The late 1980s saw the implementation of docking software packages such as TOM, integrated into FRODO, for studying protein-ligand interactions with interactive energy-minimization procedures. This allowed for the creation of models of protein-ligand complexes, followed by energy minimization treating both ligand and receptor parts as flexible units (Cambillau & Horjales, 1988).

  • Brownian Dynamics Simulation of Protein Association: The application of Brownian Dynamics (BD) to study the diffusive dynamics and interaction of proteins marked an important advancement in understanding protein-protein and protein-ligand interactions. This method assessed the influence of individual charged amino acid residues on the docking process, facilitating the study of electrostatic charge distribution in protein docking (Northrup, Luton, Boles, & Reynolds, 1988).

Evolution of Ligand-Based Drug Design

  1. Computer-aided radiopharmaceutical design by Boudreau & Efange (1992) highlights the integration of QSAR with computational methods like quantum mechanics and molecular mechanics, emphasizing the predictive power of QSAR analyses in drug design. This paper underscores the role of computational tools in enhancing the understanding and prediction of drug-radiopharmaceutical interactions (Boudreau & Efange, 1992).

  2. Strategies for Indirect Computer-Aided Drug Design by Loew, Villar, & Alkorta (1993) discusses methods used in computer-aided drug design when the structure of the target macromolecule is unknown, emphasizing the indirect characterization of ligands through QSAR and pharmacophore development. This paper provides insight into the challenges and strategies of drug design in the absence of detailed structural information about the target (Loew et al., 1993).

  3. Multivariate design and modeling in QSAR by Eriksson & Johansson (1996) explores the crucial steps in developing QSAR models, focusing on the selection of proper data analytical methods, training set design, and validation of QSAR models. This paper highlights the significance of chemometric techniques in QSAR development, relevant to both drug design and environmental sciences (Eriksson & Johansson, 1996).

  4. [Development of quantitative structure-activity relationships and computer-aided drug design] by Moriguchi (1994) reviews recent developments in QSAR and computer-aided drug design, including pattern recognition methods for analyzing structure-activity rating data and constructing predictive models for drug design. This paper emphasizes the utility of simple and fast 2D descriptors and molecular mechanical conformational analysis in the QSAR analysis (Moriguchi, 1994).

  5. Toward minimalistic modeling of oral drug absorption by Oprea & Gottfries (1999) presents a QSAR model correlating human intestinal absorption and Caco-2 cell permeability data to molecular structures. The model emphasizes a minimalistic approach to predicting oral absorption, integrating hydrophobicity and H-bonding capacity as key descriptors (Oprea & Gottfries, 1999).

Advances in pharmacophore modeling techniques: "Pharmacophore Fingerprinting. 1. Application to QSAR and Focused Library Design" by McGregor & Muskal (1999) introduced a rapid pharmacophore fingerprinting method, showcasing the methodological advancements in pharmacophore modeling during this era (McGregor & Muskal, 1999).

  1. Lead generation using pharmacophore mapping and three-dimensional database searching: application to muscarinic M(3) receptor antagonists by Marriott et al. (1999) presents an example of identifying potent novel lead compounds using pharmacophore modeling for muscarinic M(3) receptor antagonists. This study demonstrates the utility of pharmacophore models in 3D database searching and medium-throughput screening to discover compounds with desired biological activity (Marriott et al., 1999).

  2. Novel approach to predicting P450-mediated drug metabolism: development of a combined protein and pharmacophore model for CYP2D6 by de Groot et al. (1999) discusses the creation of a combined protein and pharmacophore model for cytochrome P450 2D6 (CYP2D6), integrating pharmacophore modeling, protein modeling, and molecular orbital calculations. This comprehensive approach was used to account for steric, electronic, and chemical stability properties, aligning with experimental data and site-directed mutagenesis results (de Groot et al., 1999).

  3. Conformational analysis, pharmacophore identification, and comparative molecular field analysis of ligands for the neuromodulatory sigma 3 receptor by Myers et al. (1994) carried out molecular modeling studies on various ligands showing affinity for the sigma 3 receptor. By employing pharmacophore mapping and comparative molecular field analysis (CoMFA), this study aimed to develop a ligand-binding model for the sigma 3 receptor, demonstrating the application of pharmacophore models in analyzing ligand-receptor interactions (Myers et al., 1994).

Integration of Virtual Screening and High-Throughput Screening (HTS)

  1. Integration of Virtual and High-Throughput Screening: This review highlights how high-throughput and virtual screening are complementary components of modern drug discovery, detailing various methods introduced to foster their integration. It emphasizes the potential benefits of a unified approach to biological screening in early-stage drug discovery (Bajorath, 2002).

  2. High-throughput and Virtual Screening: Core Lead Discovery Technologies Move Towards Integration: This paper describes the synergies between HTS and virtual screening (VS), discussing developments in VS technology and their potential impact on HTS, including focused screening and data mining (Good, Krystek, & Mason, 2000).

  3. Virtual Screening: An Overview: Discusses advances in combinatorial chemistry and HTS, allowing for the synthesis of large numbers of compounds and how virtual screening can reduce a huge virtual library to a manageable size for further evaluation (Walters, Stahl, & Murcko, 1998).

  4. Virtual High-Throughput in Silico Screening: This paper covers the application of in silico approaches, such as docking and alignment, to large virtual molecular databases to enrich biologically active compounds, highlighting the cost and time benefits of virtual high-throughput screening (vHTS) (Seifert, Wolf, & Vitt, 2003).

  5. Integration of Virtual Screening into the Drug Discovery Process: Reviews advances in virtual screening using docking, predictive ADME methods, and their integration with informatics and computing to enhance drug discovery processes (Chin, Chuaqui, & Singh, 2004).

Advances in Molecular Dynamics Simulations

  1. Molecular dynamics simulations of biomolecules by Geisbrecht, B., Gould, S., & Berg, J. (2002). This review highlights the importance of molecular dynamics simulations in understanding the structure and function of biological macromolecules. It discusses the transition from viewing proteins as rigid structures to recognizing their dynamic nature essential for function (Geisbrecht, Gould, & Berg, 2002).

  2. Biomolecular simulations: recent developments in force fields, simulations of enzyme catalysis, protein-ligand, protein-protein, and protein-nucleic acid noncovalent interactions by Wang, W., Donini, O., Reyes, C., & Kollman, P. (2001). This paper discusses the cornerstone of computer simulations, the force field, and its applications in understanding enzyme catalysis and macromolecular dynamics and interactions (Wang, Donini, Reyes, & Kollman, 2001).

  3. Molecular dynamics simulations of 14 HIV protease mutants in complexes with indinavir by Chen, X., Weber, I., & Harrison, R. (2004). This study utilized molecular dynamics simulations to understand the molecular mechanisms behind HIV drug resistance when interacting with the inhibitor indinavir (Chen, Weber, & Harrison, 2004).

  4. Accelerated molecular dynamics: a promising and efficient simulation method for biomolecules by Hamelberg, D., Mongan, J. T., & McCammon, J. A. (2004). The paper presents an accelerated molecular dynamics approach that allows the efficient simulation of biomolecular systems, highlighting its potential in sampling conformational space more efficiently than normal molecular dynamics simulations (Hamelberg, Mongan, & McCammon, 2004).

The Era of Big Data and Machine Learning

  1. Deep learning for big data applications in CAD and PLM: This study focuses on the applications of machine learning and deep learning in the manufacturing industry, with a case study on object recognition in heterogeneous formats using deep learning techniques (Dekhtiar et al., 2018).

  2. Machine learning on big data: Opportunities and challenges: Discusses the opportunities and challenges of machine learning (ML) in the context of big data, including model scalability and distributed computing (Zhou et al., 2017).

  3. Data Mining and Analytics in the Process Industry: Reviews applications of machine learning in the process industry for data mining and analytics, highlighting the role of ML in information extraction, data pattern recognition, and predictions (Ge et al., 2017).

  4. Machine Learning With Big Data: Challenges and Approaches: Compiles and summarizes challenges of machine learning with big data, focusing on volume, velocity, variety, or veracity, and discusses emerging ML approaches (L’Heureux et al., 2017).

  5. Big data and artificial intelligence (AI) methodologies for computer-aided drug design (CADD): Outlines the development of computational and statistical methods applying big data and AI techniques for CADD, including predictive models for ADMET properties (Lee et al., 2022).

  • Generative AI for Medicinal Chemistry: Generative AI, particularly deep learning generative models, has revolutionized the approach to de novo drug discovery by enabling the design of new medicines through creative computational methods. The application of these AI techniques requires rigorous evaluation to ascertain their real-world utility in drug discovery settings (Walters & Murcko, 2020).

  • Ligand-Based Novel Drug Discovery: Deep learning, through discriminative and generative neural network models, has become crucial for ligand-based novel drug discovery. This area covers virtual screening, neural generative models, and mutation-based structure generation, showcasing the variety and potential of deep learning in generating new molecules with desired properties (Baskin, 2020).

  • Generative Chemistry: Focused on using generative modeling to speed up the drug discovery process, this field explores cutting-edge generative architectures like recurrent neural networks, variational autoencoders, and generative adversarial networks for compound generation. These technologies are key to advancing generative chemistry and, by extension, drug discovery (Bian & Xie, 2020).

  • Deep Reinforcement Learning for Drug Design: This approach integrates generative and predictive models for the design of chemical libraries targeting specific physical and/or biological properties. It exemplifies how AI can generate targeted chemical libraries optimized for desired properties (Popova, Isayev, & Tropsha, 2017).

  • 3D Generative Models for Drug Design: Exploring the less-trodden path of 3D molecule generation, DeepLigBuilder offers a novel method for structure-based de novo drug design by generating 3D molecular structures within target binding sites, pushing the frontier in personalized medicine and targeted therapy (Li, Pei, & Lai, 2021).

  • AI in Scaffold-based Design: Generative models are evolving to allow for the incorporation of desired scaffolds directly in the generation process, enabling the creation of novel therapeutic candidates while preserving critical functional groups. This progress is pivotal for designing compounds tailored to specific biological targets, which is a cornerstone of personalized medicine (Joshi et al., 2021).

  • Hybrid Quantum Mechanics/Molecular Mechanics (QM/MM) Simulations for Drug Design: Kulkarni et al. (2021) discuss the application of hybrid QM/MM simulations in structure-based computational methods for therapeutic agent design and discovery, emphasizing their role in high-throughput screening (HTS) and computational chemistry (Kulkarni, Shah, & Vyas, 2021).

  • Scalable Molecular Dynamics with NAMD: Phillips et al. (2020) review NAMD, a program for high-performance molecular dynamics simulations that is versatile for simulations in various thermodynamic ensembles, enabling large-scale simulations on both CPU and GPU architectures (Phillips et al., 2020).

  • Emerging Quantum Computing Algorithms for Quantum Chemistry: Motta and Rice (2021) provide an introduction to emerging algorithms for quantum chemistry simulations on digital quantum computers, focusing on their applications to molecular systems' electronic structure (Motta & Rice, 2021).

  • Exploiting Chemistry for Quantum Information Science: Wasielewski et al. (2020) explore how molecular systems' quantum properties can advance quantum computing, communication, and sensing, highlighting the role of chemistry in the second quantum revolution (Wasielewski et al., 2020).

  • QuantumATK: Integrated Platform for Electronic and Atomic-scale Modelling: Smidstrup et al. (2019) overview QuantumATK, an integrated set of tools for atomic-scale modeling, including electronic-structure calculations and molecular dynamics, supporting various simulation methods (Smidstrup et al., 2019).

  • Full Quantum Eigensolver for Quantum Chemistry Simulations: Wei, Li, and Long (2019) propose a full quantum eigensolver (FQE) algorithm for calculating molecular ground energies and electronic structures on quantum computers, emphasizing faster convergence without a classical optimizer (Wei, Li, & Long, 2019).

  • Recent Advances in First-Principles Based Molecular Dynamics: Mouvet et al. (2022) discuss advances in first-principles molecular dynamics (FPMD) and quantum mechanical-molecular mechanical (QM/MM) extensions for simulating a broad variety of systems (Mouvet et al., 2022).

  • Machine Learning for Molecular Simulation: Noé et al. (2019) review machine learning methods for molecular simulation, focusing on neural networks for predicting quantum-mechanical energies and forces, and their application in molecular dynamics, free energy surfaces, and kinetics (Noé, Tkatchenko, Müller, & Clementi, 2019).