In Silico Docking: Protocols for Computational Exploration of Molecular Interactions

Abstract

In computational chemistry and drug development, in silico docking has become an indispensable tool for investigating the molecular interactions between ligands and receptors. The procedures and approaches used in in-silico docking to decipher the complex dynamics of molecular binding processes are highlighted in this chapter. The first section of the chapter explains the basic ideas behind molecular docking, focusing on the function of scoring functions and algorithms in ligand-receptor interaction prediction. The benefits and drawbacks of several docking techniques—such as flexible docking, rigid-body docking, and other docking methods—are thoroughly covered. In addition, the challenges associated with conformational flexibility, solvent effects, and ligand desolvation that arise during in-silico docking are explored. Molecular dynamics simulations and ensemble docking techniques are investigated as ways to improve the precision and dependability of docking predictions. Furthermore, using in silico docking in virtual screening, structure-based drug design, and drug discovery highlights how important it is to speed up the drug development process and reduce experimental expenses. A thorough review of in silico docking techniques is given in this chapter, along with an examination of its methodological complexities, theoretical underpinnings, and real-world uses in drug discovery and computational chemistry.

1. Introduction

The completion of the human genome project has led to an increasing number of new therapeutic targets for drug discovery. Simultaneously, advancements in high-throughput protein purification, crystallography, and nuclear magnetic resonance spectroscopy techniques have provided detailed insights into proteins and protein-ligand complexes. These breakthroughs have enabled computational strategies to play a pivotal role in various aspects of drug discovery today [1, 2, 3, 4, 5], including virtual screening (VS) techniques [6] for hit identification and methods for lead optimization. Compared to traditional experimental high-throughput screening (HTS), virtual screening (VS) represents a more direct and rational approach to drug discovery, offering the advantages of low-cost and effective screening [7, 8, 9]. VS can be categorized into ligand-based and structure-based methods. Ligand-based methods, such as pharmacophore modeling and quantitative structure-activity relationship (QSAR) methods, are employed when a set of active ligand molecules is known, and little or no structural information is available for the targets [10]. On the other hand, structure-based methods rely on information derived from the 3D structure of a target of interest, allowing the ranking of molecule databases based on the structural and electronic complementarity of ligands to a given target [5].

In this context, molecular docking stands out as one of the most popular and successful structure-based in silico methods. It aids in predicting interactions between molecules and biological targets [5]. The process generally involves predicting the molecular orientation of a ligand within a receptor and then assessing their complementarity using a scoring function [5]. Since its initial emergence in the mid-1970s, docking has proven to be a crucial tool for comprehending how chemical compounds interact with their molecular targets, aiding in drug discovery and development. In fact, the number of studies reporting both the utilization of molecular docking to identify structural determinants essential for efficient ligand-receptor binding and the advancement of more precise docking methods has significantly increased since its inception [5, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]. One of the earliest and most noteworthy studies on the application of docking in drug discovery and biology was conducted by Kuntz et al. in the early 1980s [10]. In this study, the authors presented a computational method that allowed the exploration of geometrically feasible ligand-receptor alignments for known heme-myoglobin/metmyoglobin and thyroxine/prealbumin structures [10]. While not the first study to employ docking for predicting potential conformations of molecular complexes [12], it marked the first use of a simplified function containing solely “hard sphere repulsions” and “hydrogen bonding” terms to describe protein-ligand interactions, which differed significantly from previous studies [12, 14, 15, 24]. Additionally, the authors were the first to consider the receptor as a solid rigid body with a binding site constituted by “pockets.” Notably, the method in this study successfully predicted structures close to those of previously reported X-ray complexes and identified protein conformations suitable for energy refinement and potential ligand design [10].

However, the application of docking in drug design is constrained to biological targets with known crystal structures. Various approaches have been employed to overcome this limitation, such as constructing homology models derived from structural templates with highly homologous sequences when 3D structures are unavailable. Furthermore, these methods can be used in conjunction with molecular dynamics (MD) to validate and refine in silico modeled complexes [25, 26, 27]. Nevertheless, recent advancements in structural biology and crystal structure determination, progressively increasing access to experimentally derived ligand-target complexes [28, 29, 30, 31], will undoubtedly alleviate this issue. In silico strategies, including molecular dynamics, have extensively explored the conformational space of investigated targets, ligands, and ligand-target complexes, providing a more comprehensive understanding of the dynamic behavior of ligand-target complexes and refining docking results [32, 33]. More rigorous virtual screening methodologies have also been developed to enhance docking-based predictions of ligand-target complexes [34, 35, 36, 37, 38]. Indeed, these post-docking refinement and rescoring methods hold great significance in drug discovery, as they typically yield higher hit rates in virtual screening campaigns and demonstrate improved correlation with experimental data [36, 38]. Several reviews have addressed the role and applications of docking, exploring the potential it holds in drug design and development [39, 40, 41, 42]. It is important to note that the uses and applications of docking have evolved since its inception. While initially designed to investigate molecular recognition between large and small molecules, docking is now extensively employed in various aspects of drug discovery programs. These include hit identification and optimization, drug repositioning, a posteriori target identification (reverse screening), multi-target ligand design, and repositioning (see Figure 1) [43, 44, 45, 46, 47, 48, 49, 50].

Figure 1.

Main applications of molecular docking in current drug discovery. Molecular docking is currently employed to help rationalize ligands activity toward a target of interest and to perform structure-based virtual screening campaigns, similarly to as when it was first developed. Besides these applications, it can also be used to identify series of targets for which the ligands present good complementarity (target fishing and profiling), some of them being potentially responsible for unexpected drug adverse reactions (off-targets prediction). Moreover, docking is also currently employed for the identification of ligands that simultaneously bind to a pool of selected targets of interest (polypharmacology) and for identifying novel uses for chemical compounds with already optimized safety profiles (drug repositioning).

Furthermore, docking facilitates the comprehension of relationships between different molecular targets associated with a specific disease, a crucial aspect for polypharmacology [51] and modern drug discovery in general.

In this chapter, we will discuss how various docking methods have been used to help assisting drug discovery tasks, giving particular emphasis on choosing the right software and docking protocols including challenges and limitations.

2. Fundamentals of molecular docking

2.1 Molecular docking defined

Molecular docking serves as a crucial tool in both structural molecular biology and computer-assisted drug design. Its primary objective is to forecast the primary binding mode(s) between a ligand and a protein with a known three-dimensional structure. The success of docking methods relies on their ability to effectively explore high-dimensional spaces and employ a scoring function that accurately prioritizes potential dockings. Utilizing docking enables virtual screening of extensive compound libraries, facilitating result ranking and proposing structural hypotheses regarding how ligands hinder the target—making it invaluable for lead optimization. It is essential to emphasize that the configuration of input structures for docking is as significant as the docking process itself [52]. Furthermore, analyzing results from stochastic search methods may sometimes pose challenges. In essence, molecular docking, a computational technique, predicts ligand-receptor binding affinity and has evolved into a formidable asset in drug development, with potential applications in nutraceutical research [53].

2.1.1 Explanation of molecular docking in computational biology

Completing the human genome project has uncovered numerous therapeutic targets for drug discovery. Advances in protein purification and structural analysis techniques, such as crystallography and nuclear magnetic resonance spectroscopy, have facilitated the integration of computational strategies in drug discovery. Virtual screening, a cost-effective and efficient approach, includes ligand-based and structure-based methods. Ligand-based methods, like pharmacophore modeling, are used when target structural information is limited [54] Molecular docking, a prominent method in structure-based drug design, models atomic-level interactions between small molecules and proteins, aiding in understanding biochemical processes. Efficient docking is achieved when the binding site is known through prior knowledge or identification using cavity detection programs. Early theories, like Fischer’s lock-and-key and Koshland’s induced fit, influenced molecular docking approaches [20]. Despite computer resource limitations, flexible ligand and rigid receptor docking remain popular, with ongoing efforts to address receptor flexibility challenges, including the proposed Local Move Monte Carlo (LMMC) approach [10].

2.1.2 Historical perspective

Molecular docking is a computer method utilized in drug development and structural biology. It includes predicting a molecule’s preferred shape or orientation to form a stable complex bound to a target receptor. The historical perspective of this approach spans multiple decades:

1. Emergence of Computational Chemistry (1960s–1970s): the 1960s and 1970s saw the emergence of computational chemistry. Molecular docking originated in the early years of computational chemistry when scientists started modeling and simulating molecular interactions using computers. The development of algorithms and the accessibility of computers provided the groundwork for the use of computational methods in the study of molecular interactions and structures [55]

2. Invention of Molecular Mechanics (1970s–1980s): more accurate simulations of molecular interactions were made possible by the invention of molecular mechanics force fields, which characterize the energy and forces in a molecular system. These force fields were first used by scientists to simulate how molecules would behave in three dimensions [56]

3. The Advent of Molecular Dynamics (1970s–1980s): this dynamic perspective on molecular interactions was made possible by molecular dynamics simulations, which imitate the real-time movements of atoms and molecules. Researchers were able to watch how molecules behaved and bound to receptors over time thanks to this technology [55, 56]

4. Pioneering Work (1980s–1990s) on Docking Algorithms: in the 1980s and 1990s, few researchers contributed to the creation of the first molecular docking algorithms. The initial docking algorithms relied on basic geometric and empirical scoring functions to anticipate the binding affinity and alignment of ligands within the target protein’s binding region [57].

5. Scoring Function Advancements (1990s–2000s): with the advent of increasingly complex scoring functions, docking algorithms saw notable advancements in the 1990s and 2000s. The purpose of these scoring functions was to differentiate between genuine binding poses and false positives, as well as to better anticipate the binding free energy [55].

6. Use in Drug Discovery (2000s–Present): in the pharmaceutical sector, molecular docking is now a crucial component of drug discovery workflows. Large compound libraries are screened using it to find possible therapeutic candidates, forecast binding affinities, and enhance lead compounds.

7. Machine Learning Advancements (2010s-Present): the accuracy of molecular docking predictions has been significantly improved in recent years by including machine learning techniques, such as deep learning. These methods enhance our comprehension of intricate molecular interactions by utilizing extensive datasets and advanced algorithms [58].

The advancement of computational methods and the incorporation of experimental data to enhance comprehension and forecast molecular interactions for diverse uses, most notably drug discovery, are reflected in the historical development of molecular docking. Artificial intelligence (AI) has been applied to various docking systems in different domains. In the field of cross-docking systems, AI-based techniques have been explored to solve cross-docking problems, and new potential uses of AI-based techniques have been identified [59]. In the context of subsea homing and docking systems for autonomous underwater vehicles (AUVs), an AI-enabled short-range AUV electromagnetic homing guidance system (EMHGS) has been developed using supervised and unsupervised machine learning algorithms [60]. In the domain of protein-ligand docking, high-speed ML models have been used to accelerate the scoring of compounds, and AI-based models have been utilized as a pre-filter to improve the speed and accuracy of virtual screening [61]. In the field of on-orbit servicing (OOS), an AI-powered navigation algorithm has been developed for autonomous docking and refueling of spacecraft [62]. In the domain of underwater docking, deep learning techniques, including knowledge distillation and synthetic data generation, have been used to efficiently train convolutional neural networks (CNNs) for docking station detection [63].

2.2 Biological relevance

Molecular docking is a computational method employed to identify the correct binding pose of a protein-ligand complex and assess its strength through various scoring functions, ranking the best pose for each molecule [10]. The technique aims to fit a ligand into the target protein’s binding site by optimizing variables like hydrophobic, steric, and electrostatic complementarity, estimating their binding free energy. Successful docking relies on high-resolution structures, obtained through X-ray crystallography, NMR, or homology modeling, with a known binding site.

The Protein Data Bank (PDB) contains thousands of experimentally determined 3D structures of proteins. X-ray and NMR machines enable the detailed resolution of molecular interactions, crucial for recognizing key residues, analyzing interaction forces, and energetics, and understanding structural fits. Some protein structures are challenging to determine experimentally, leading to their absence in the PDB. Yet, advancements in bioinformatics, like homology modeling, enable the prediction of unavailable protein structures using templates with similar residues. Servers include Swiss Model, Phyre [58] i-TASSER, HH Pred, PSI-Pred, Robetta, Raptor, and the groundbreaking AlphaFold by DeepMind [58]. It has revolutionized computational structural biology in 2021, achieving over 92% accuracy in 3D structural prediction [64]. AlphaFold and RoseTTAFold, another accurate model, hold promise for transforming in silico drug discovery in the twenty-first century, with RoseTTAFold excelling in predicting complex biological assemblies like protein-protein interactions [64].

2.3 Types of molecular docking

Molecular docking is a valuable tool in drug discovery research because it makes target selection, validation, virtual lead identification screening, and lead optimization easier. Drugs are primarily organic tiny molecules; however, targets might be either protein or DNA depending on the type of sickness of concern. To accomplish these objectives, many molecular docking approaches are used, such as protein-protein, protein-DNA, protein-small molecule, and DNA-small molecule. Majorly docking can be classified into two categories:

• Flexible docking: addressing flexibility in molecular docking poses a significant challenge in cell biology research. The flexible docking process comprises four main stages. Firstly, in the preprocessing stage, proteins undergo analysis to define their conformational space, allowing the generation of discrete conformations. This ensemble is used in cross-docking, simulating the conformational selection model. Hinge locations may also be identified, leading to the separation of rigid and flexible parts for individual docking [8]. The second stage involves rigid docking, aiming to produce solution candidates with at least one near-native structure, permitting some steric clashes. The subsequent refinement stage models induced fit, optimizing each candidate through small backbone and side-chain movements and rigid-body adjustments. This optimization is performed in successive steps for side-chain conformations, backbone structure, and rigid-body orientation. The refined structures exhibit improved binding energy and minimal steric clashes. The final scoring stage assesses and ranks candidate solutions based on parameters like binding energy, agreement with known binding sites, deformation energy of flexible proteins, and the presence of energy funnels [65].

• Rigid docking: rigid docking is a stage in molecular docking where the docking procedure is conducted without allowing significant flexibility or movement in the structures of the molecules being docked. In other words, the interacting molecules, typically a ligand and a receptor (e.g., protein), are treated as rigid bodies during the docking simulation. This approach assumes that the overall structure of the molecules remains fixed, and only translational and rotational degrees of freedom are considered in the docking process [10]. During rigid docking, the goal is to generate a set of potential docking poses for the ligand within the binding site of the receptor. This stage aims to identify configurations where the ligand and receptor interact favorably, forming a stable complex. While the rigid docking stage allows for some steric clashes, it does not consider significant conformational changes or flexibility in the molecular structures [66]. Rigid docking is often followed by a refinement stage, where more detailed adjustments, including flexible movements of side chains or larger conformational changes, may be considered to improve the accuracy of the predicted binding poses. Employing rigid or flexible docking depends on the specific goals of the docking study and the nature of the interactions being investigated [67].

2.4 Other variations and hybrid methods

In molecular docking, hybrid methods pertain to the amalgamation of many computational methodologies or approaches to enhance the precision and dependability of forecasts. These hybrid approaches make up for the shortcomings of individual techniques by combining the advantages of several methodologies. The following are a few typical hybrid molecular docking technique types:

2.4.1 Hybrid techniques based on structure and ligands

Pharmacophore-Based Docking: combines structure-based docking and ligand-based pharmacophore modeling. New compounds are docked using pharmacophore properties that come from well-known ligands [5].

Shape-Based Docking: integrates structure-based docking with knowledge of the shape of the ligand. The shape similarity between ligands and the target binding site is considered during the docking process [6].

2.4.2 Hybrid docking methods and molecular dynamics (MD)

The docking procedure is guided by data obtained from molecular dynamics simulations. The accuracy of docking poses can be improved, and conformational changes observed during MD simulations can be used as guidance using molecular dynamics simulations on anticipated docking poses, MD-refined docking involves fine-tuning and optimizing the complicated structures of ligands and receptors [6].

2.4.3 Docking and quantum mechanics/molecular mechanics (QM/MM)

QM/MM-docking combines molecular mechanics for the remainder of the system with quantum mechanics for precisely handling sections (such as the active site). Studying processes catalyzed by enzymes is an excellent use for this hybrid approach.

2.4.4 Docking and machine learning: ML-based scoring functions

Combines machine learning models to enhance the scoring functions employed in docking investigations. With the help of experimental data, machine learning algorithms can more precisely estimate binding affinities. ML-enhanced pose prediction: combines machine learning techniques with docking algorithms to improve pose prediction accuracy, especially for flexible ligands and dynamic binding sites.

2.4.5 Ensemble docking

Multiple receptor conformations: involves using an ensemble of receptor structures (e.g., from different conformations or experimental structures) during the docking process. This approach considers the inherent flexibility of the receptor.

2.4.6 Hybrid ligand screening

It improves the discovery of putative ligands from extensive compound libraries by fusing structure-based docking with ligand-based virtual screening.

2.4.7 Docking and free energy calculations

To estimate the binding free energy of ligands more precisely, free energy perturbation (FEP) docking integrates docking with free energy calculations.

These hybrid approaches seek to overcome issues with molecular docking scoring accuracy, ligand flexibility, and receptor flexibility. Enhancing the efficacy of molecular docking in drug discovery and structural biology, researchers can attain more dependable predictions of ligand binding mechanisms and affinities by merging complementing techniques.

3. Computational tools and software

A vital tool in the drug-discovery process, docking software helps scientists understand how small compounds interact with proteins of interest, leading to the creation of novel therapeutic medicines. To predict the ideal binding configuration. Three areas use computational tools: virtual screening to find compounds with high binding potential through large databases, lead optimization to improve lead compound properties, and drug discovery to find potential drug candidates by predicting their binding affinity to target proteins [68, 69].

3.1 Overview of docking software

Docking software predicts the binding mechanism and affinity of small compounds with a target biomolecule, usually a protein, and is therefore essential to computational chemistry and drug discovery [70].

3.2 AutoDock Vina

AutoDock Vina is a molecular docking software used in computational chemistry to predict the binding modes of small molecules or ligands with a target protein. It is an improved version of the original AutoDock software and is widely used for virtual screening and drug discovery studies [70].

Key features of AutoDock Vina include:

Scoring function: AutoDock Vina uses an advanced scoring function to evaluate the binding affinity between a ligand and a target protein. The scoring function takes into account various energy terms including steric clashes, hydrogen bonding, and electrostatic interactions.

Search algorithm: it employs a global optimization algorithm called iterated local search (ILS) to explore the conformational space efficiency and find energetically favorable binding modes.

Flexible ligand and receptor handling: AutoDock Vina allows for the flexibility of both ligands and receptors. This is important as proteins and ligands can undergo conformational changes upon binding.

User-friendly interface: while AutoDock Vina itself is a command line tool, there are graphical user interfaces (GUIs) available that provide a more user-friendly environment for setting up and running docking simulations.

Open source: AutoDock Vina is open-source software, making it freely available for academic and research purposes. Users can access the source code, modify it, and contribute to its development.

Integration with visualization tools: results from AutoDock Vina can be visualized using molecular visualization tools, such as PyMOL or visual molecular dynamics (VMD), to analyze the binding modes and interactions between ligands and proteins. It is important to note that AutoDock Vina is one of many molecular docking tools available, and the choice of which tool to use depends on the specific requirements of the research or virtual screening project. Users are encouraged to refer to the official documentation and publications associated with AutoDock Vina for detailed information on its usage and capabilities [71].

3.2.1 Glide

A Schrodinger company known for its suite of computational chemistry and molecular modeling software does offer a docking program called Glide [19]. Glide is a molecular docking program used for predicting the binding modes and affinity of small molecules to target proteins. It is widely used in drug discovery and structure-based drug design. Glide is the primary industrial solution for reliable ligand-receptor docking. It improves and speeds up structure-based drug design for a variety of uses, such as interactive 3D molecular design, virtual screening, and binding mode prediction. It is easy to understand that a guided graphical user interface makes it simple to develop and test docking models and also achieve high enrichment for small compounds, peptides, and macrocycles among a wide variety of receptor types. Additionally, it offers simple, constraint-free docking computations that satisfy the criteria and desired chemical space as found in experiments. Schrödinger programs can be executed via a command line or an interface with graphics. The working directory is the location to which the software writes input and output files. The working directory for the task is the location from which you launch the application when using the command line.

The software package developed by Schrodinger utilizes several methods and methodologies to facilitate activities associated with drug development, molecular modeling, and computational chemistry. The Glide algorithm, which is utilized for molecular docking, is one of the well-known algorithms in Schrodinger’s program. The first stage of Glide, High Throughput Virtual Screening (HTVS), is intended for the quick screening of many ligands.

Standard Precision (SP): using a more precise scoring function and considering a smaller subset of ligands, this phase improves upon the findings from HTVS.

Extra Precision (XP): this Glide phase is the most precise and computationally intensive. Regarding ligand-receptor interaction prediction, it offers the highest degree of accuracy.

Schrodinger’s software suite includes Glide and several other algorithms and tools for various applications, including pharmacophore modeling, quantum mechanics computations, and molecular dynamics simulations [72].

3.2.2 Genetic optimization for ligand docking (GOLD)

A genetic technique called Genetic Optimization for Ligand Docking (GOLD) is used to dock flexible ligands into protein binding sites [19, 20]. GOLD has undergone comprehensive testing and demonstrated outstanding performance in virtual screening and strong outcomes in pose prediction. GOLD optimizes ligand placement by exploring the conformational space of the ligands within the binding site using a genetic algorithm. The scoring function in GOLD is critical for measuring the fitness of different ligand conformations created during the docking process. Typical terms and components relevant to GOLD scoring functions in molecular docking software are as follows:

Vander Waals interactions explain the forces that attract and repel atoms. Electrostatic interactions consider the interactions between charged particles, such as those between positively and negatively charged atoms.

Hydrogen bonding: the creation of hydrogen bonds between the ligand and the protein may be considered a scoring function.

Solvation energy refers to the energy needed to dissolve the binding site and the ligand.

Entropic contributions: after binding, entropy changes may also be considered by scoring functions.

Flexibility: by permitting ligand flexibility or considering protein-induced-fit effects, several scoring methods consider both the ligand and the protein’s flexibility [73].

3.2.3 Moledock/Molegro

Molegro Virtual Docker is an integrated platform for predicting protein-ligand interactions, from preparing the compounds to identifying possible target protein binding sites and forecasting ligand binding modalities, Molegro Virtual Docker manages every step of the docking process [74].

Molegro Virtual Docker is a user interface that prioritizes productivity and usability while providing high-quality docking based on a unique optimization technique.

High docking accuracy: it has been demonstrated that the docking engine can accurately recognize binding modes. It has been demonstrated that Molegro Virtual Docker performs better than other docking programs when it comes to identifying the proper binding modes.

Easy-to-use interface: the integrated wizards simplify the process of setting up and executing docking runs for the user. Sophisticated visualization and analysis tools are provided to study ligand-receptor interactions and optimize found docking solutions [75].

3.2.4 FlexX

One important tool for assessing and ranking the possible binding positions of ligands within a target binding site is the scoring function included in molecular docking software, such as FlexX [76, 77]. To assist researchers in determining the most likely and biologically significant binding modes, the scoring function attempts to estimate the binding affinity between the ligand and the receptor. One important tool for assessing and ranking the possible binding positions of ligands within a target binding site is the scoring function included in FlexX. To assist researchers in determining the most likely and biologically significant binding modes, the scoring function attempts to estimate the binding affinity between the ligand and the receptor. FlexX evaluates a docking pose’s quality using a scoring algorithm that considers several variables [78].

Ligand-receptor interaction energy: this metric assesses how strongly a ligand and receptor interact. Van der Waals forces, hydrogen bonds, and electrostatic interactions are examples of this.

Desolvation energy: the cost of desolvating the ligand and receptor as they assemble in the binding site is expressed in terms of desolvation energy.

Entropy-related terms: certain scoring functions take conformational entropy or loss of freedom into account when calculating scores.

Hydrophobic interactions: the scoring function may additionally take hydrophobic contact contributions into account.

Additional structural and energetic features: additional terminology about receptor flexibility, ligand flexibility, and other structural or energetic qualities may be provided, depending on the particular program.

3.2.5 Surflex-Dock

A molecular docking tool called Surflex-Dock is used to forecast the affinities and binding patterns of ligands to protein targets [78]. The Hammerhead scoring function, which considers both form complementarity and electrostatic interactions between ligands and proteins, serves as the foundation for the scoring system in Surflex-Dock. The fitness of a ligand posture inside a protein binding site is assessed using the scoring function [79].

Shape complementarity (Sshape): measure how well the ligand fits into the protein’s binding site according to shape.

Chemical score (Schem): analyze the chemical complementarity of the ligand and protein using the chemical score. Ligands with advantageous chemical interactions obtain better scores.

Solvation (Ssolve): explains the change in salvation energy upon ligand binding and encourages ligands that displace water molecules from the binding site.

Torsional Strain (Stors): penalizes strained torsional angles in the ligand. Lower scores are assigned to ligand poses with unfavorable torsional geometry.

3.2.6 ICM-pro

ICM-Pro is a molecular docking tool, and user guides and documentation are available from the program’s creators or the official Web site. With direct access to sequence and structural databases, the docking software ICM-Pro expands the capabilities available to biologists and chemists. User can perform the following tasks: generate surfaces, compute electrostatics, introduce mutations, analyze sequences and alignments, look at protein structure, look at pockets and bound ligands and medications, and dock tiny molecules and proteins with one another. Determine the locations of ligand binding sites and create ligands with the Interactive Ligand Editor (Figure 2) [71, 75].

Figure 2.

Different docking tools available.

3.3 Choosing the right software

Choosing the molecular docking software depends upon the type of work or research you are doing, your budget, your level of experience, and your unique needs will all play a role in selecting the best molecular software. When choosing molecular software, keep the following important factors in mind.

3.3.1 Considerations for selecting docking software

• User interface (UI) and user-friendliness

• Integration and compatibility

• Compliance and data security

• Trial period

• Cost and licensing

• Compliance and data security

• Features and functionality.

A user interface that is easy to use is essential, especially if your team is diverse and comprises members with varying skill sets. Look for software that has a simple user interface, clear navigation, and comprehensive help. Check to see if the software works with the operating system you are using Windows, macOS, or Linux. Consider the program’s compatibility with databases and other applications that you may find yourself using. If you work with sensitive data or are required to follow particular instructions, make sure the program conforms with security standards and legislation. Use trial copies or demos of the program to determine how well it fits into your particular workflow before making a final choice. Regardless matter whether the product is open source, commercial, or freeware, find out the license structure. Keeping in mind your budgetary limitations, select a program that fits within what you can afford. If the application must follow certain criteria or you work with important information, make sure it conforms with security norms and regulations. Verify that the program includes the exact features and capabilities you need for your research or task. This could involve docking investigations, dynamics simulations, simulation, visualization, and molecular modeling, among other things [80, 81].

3.3.2 Preparation of ligands and receptors

This section will guide you through the essential steps involved in preparing receptors and ligands for molecular docking experiments, ensuring accurate and reliable results in computational studies.

3.3.3 Preparation of ligand

A key initial step in gaining significant and physiologically relevant information from molecular docking studies is to ensure the structural quality of ligands. To improve the accuracy of the docking predictions, researchers should carefully validate and optimize ligand structures.

3.3.3.1 Retrieval of structure

The three-dimensional structures of ligand obtained from the software like ChemDraw or Marvin Sketch, or you may utilize chemical databases like PubChem and ChEMBL.

3.3.3.2 Conversion into desired file format

Convert ligand structures into a format that docking software recognizes. Protein Data Bank (PDB) and molecular 2 (mol2) formats are common formats.

3.3.3.3 Minimizing energy

Minimize the ligands’ energy to get them into stable, low-energy conformations. This stage contributes to increasing the correctness of the docking findings. One can utilize programs such as PyMOL, Open Babel, or Avogadro.

3.3.3.4 Protonation and ionization

Reduce the ligands’ energy and transform them into stable, low-energy conformations. This step helps to improve the accuracy of the docking findings. One can use applications like Avogadro, PyMOL, and Open Babel.

3.3.4 Preparation of receptors

There are multiple phases involved in the production of receptors in the context of molecular docking to make sure that the protein structure is appropriate for docking investigations. A computer forecasts the manner and affinity of a ligand’s (small molecule’s) binding to a target protein. The following are general instructions for setting up receptors in docking software, along with links to other resources:

3.3.4.1 Protein structure retrieval

Acquire the receptor protein’s three-dimensional structure. It is possible to employ homology modeling, NMR spectroscopy, or X-ray crystallography.

3.3.4.2 File format conversion

The second step is to convert the protein structure into a file format that docking software can read. PDB format is widely utilized.

3.3.4.3 Removal of water and non-biological molecules

Remove the protein structure of any water molecules and non-biological molecules (such as ions or co-crystallized ligands). Chimera and PyMOL are two tools that can help with this process.

3.3.4.4 Add hydrogen

Verify that the hydrogen atoms are positioned correctly in the protein structure. Accurate docking predictions depend on this. To add hydrogen atoms, using programs such as PDB2PQR or Reduce.

3.3.4.5 Energy minimization

To reduce any steric conflicts and improve the geometry, minimize the energy in the protein structure. Software for molecular dynamics or programs like CHARMM or GROMACS can be utilized.

3.3.4.6 Active site positioning

Determine and describe the protein’s binding pocket, also known as the active site, where the ligand is predicted to bind. This may be supported by computational predictions, literature, or experimental results.

3.3.5 Importance of structural quality

In molecular docking investigations, the structural quality of ligands is important since it can greatly affect the precision and dependability of the docking data. The following, along with relevant references, are some salient points highlighting the significance of ligand structural quality in molecular docking:

3.3.5.1 The accuracy of binding prediction

The importance of precisely predicting binding conformations and energies is accurate ligand structures are required.

3.3.5.2 Energy minimization and conformational sampling

Improved ligand architectures enhance reliability in energy minimization and improved conformational sequencing during the docking procedure.

3.3.5.3 Handling tautomers and ionization states

Ligand structures must accurately depict the relevant ionization and tautomeric states in the context of the experiment.

3.3.5.4 Handling ionization states and tautomers

Ligand structures ought to faithfully depict the pertinent ionization and tautomeric states in the context of the experiment.

3.3.5.5 The ionization states and pH sensitivity

Based on the experiment’s results, pH should be considered, as ligand protonation states impact binding affinity.

3.3.5.6 Handling ligand flexibility

Accurate ligand structures are essential for managing ligand flexibility, an important factor in several drug-receptor interactions.

3.3.5.7 Docking software considerations

Different ligand structures may be suggested or required by different docking software.

3.3.5.8 Validation and benchmarking

Assessing experimental findings against ligand structures validates them and improves the validity of docking investigations (Figure 3).

Figure 3.

Process of molecular docking.

4. Docking protocols

4.1 Pre-docking steps

It is essential to synthesize and purify ligand and receptor molecules before using them in different biochemical and biophysical research projects. Small chemical molecules called ligands attach to particular receptors with preference. Many biological functions, such as signal transduction, cellular communication, and therapeutic efficacy, depend on this interaction between ligands and receptors. The ligand structure can be obtained from databases like ZINC [77] and Pub Chem (https://pubchem.ncbi.nlm.nih.gov/) or sketched using the Chem sketch tool. The ligand can be screened using the Lipinsky Rule of Five as a pre-docking phase for ligand preparation [82].

The Protein Data Bank (PDB) has the protein’s three-dimensional structure available for download. This recovered structure needs to go through pre-processing after that. According to the given parameters, this pre-processing entails, among other things, eliminating water molecules from the protein cavity, stabilizing charges, adding missing residues, and producing side chains [83].

4.1.1 Grid generation

The area surrounding a protein molecule’s binding site in three dimensions is commonly referred to as a grid within molecular docking processes. Various tools, such as CASTp (http://sts.bioengr.uic.edu/castp/) and LigASite A (http://www.bigre.ulb.ac.be/Users/benoit/LigASite/index.php?home), are available for predicting an active site. This grid, divided into evenly spaced points around the active site, essentially represents a discretized version of the binding site. Each point on the grid represents a potential location for a ligand molecule to bind to the protein. Assessing the protein’s active site is essential. While the receptor may possess multiple active sites, only the one presenting the highest risk must be prioritized for selection.

4.2 Docking algorithms

Docking algorithms play a crucial role in drug discovery and molecular design by predicting how small molecules (ligands) bind to larger molecules (proteins or receptors). The docking algorithms aim to determine which conformation is the most stable and energetically favorable and what forms the ligand attached to the protein could take.

4.2.1 Different docking approach

• Shape-based docking: focuses on the binding site’s and ligand’s geometric compatibility. Examples of software falling in this category include DOCK, which employs geometric matching algorithms to identify poses with optimal shape complementarity between ligand and receptor. FTDOCK utilizes the fast Fourier transform to compare the shapes of ligands and receptors efficiently. ZDOCK leverages molecular surface patches to rapidly assess geometric compatibility.

• Docking with energy grids: potential energy at each location in a discretized binding site grid is calculated throughout the docking process using energy grids. Examples of software falling in this category include AutoDock, which calculates intermolecular interaction energies (e.g., van der Waals, electrostatic) on a pre-calculated grid around the binding site. Glide [67] employs a grid-based scoring function combined with ligand conformational sampling to identify energetically favorable poses. GOLD [10] utilizes a pre-calculated potential field grid to estimate interaction energies and guide ligand placement.

• Knowledge-based docking: makes more precise predictions using data from known protein-ligand complexes. Examples of software in this category include DFIRE, scores docked poses based on their similarity to known protein-ligand complexes extracted from structural databases. ICM-Pro leverages knowledge-based potentials derived from protein-ligand interfaces to rank potential binding modes. RosettaDock [84] integrates structural information from protein-protein interfaces to evaluate docked poses (Table 1).

S. no	Algorithm	Software	Principle	Scoring function	References
1.	Lamarckian Genetic Algorithm (LGA)	AutoDock and AutoDock Vina	LGA combines aspects of genetic algorithms with local search methods. It evolves a population of candidate solutions and applies local search operators to improve solution quality.	Empirical scoring functions based on intermolecular interactions such as van der Waals forces, electrostatic interactions, and hydrogen bonding.	[56]
2.	Genetic Algorithm (GA)	GOLD and AutoDock	GA is a search heuristic inspired by natural selection and genetics principles. It maintains a population of candidate solutions and applies genetic operators such as selection, crossover, and mutation to evolve toward optimal solutions.	GA utilizes empirical scoring functions to evaluate the fitness of candidate solutions based on intermolecular interactions.	[10, 56]
3.	Differential Evolution Algorithm (DE)	AutoDock and AutoDock Vina	DE is a population-based optimization algorithm that iteratively improves candidate solutions by combining differences between randomly selected solutions.	Employs scoring functions that evaluate the energy of the docking complex, considering factors like steric clashes, hydrogen bonding, and electrostatic interactions.	[56]
4	Monte Carlo (MC) methods	Rosetta, FlexX	MC methods use random sampling techniques to estimate properties of a system by generating a large number of random configurations and averaging their contributions.	MC methods utilize scoring functions that assess the energy of the system based on molecular mechanics force fields or empirical potentials.	[84]
5.; 5	Ant Colony Optimization (ACO)	SwarmDock	The foraging behavior of ants inspires ACO. It utilizes pheromone trails left by ants to guide the search for optimal solutions in a graph-based problem space.	ACO relies on scoring functions that evaluate the fitness of candidate solutions by considering the complementarity of molecular surfaces and the strength of intermolecular interactions.	[85]
6.	Particle Swarm Optimization (PSO)	eHiTS and SwarmDock	PSO is inspired by the social behavior of bird flocking and fish schooling. It maintains a population of candidate solutions (particles) that move through the search space based on their best position and the global best position the swarm finds.	PSO utilizes scoring functions similar to those used in other optimization algorithms, assessing the fitness of candidate solutions based on the energy of the docking complex and the quality of molecular interactions.	[85]
7.	Tabu Search (TS)	FlexX	TS is a metaheuristic search method that iteratively explores the solution space by maintaining a tabu list of recently visited solutions and applying neighborhood search techniques to escape local optima.	TS employs scoring functions that evaluate the energy and stability of candidate docking configurations, often based on empirical potentials or molecular mechanics force fields.	[86]
8.	Simulated Annealing (SA)	Rosetta and FlexX	The annealing process inspires SA in metallurgy. It probabilistically accepts worse solutions in the early stages of optimization and gradually decreases the probability as the algorithm progresses.	SA utilizes scoring functions that evaluate the energy of the system and guide the search for low-energy conformations based on molecular mechanics force fields or empirical potentials.	[84, 86]
9.	Iterated Local Search (ILS)	PatchDock, SwarmDock	ILS iteratively explores the solution space by perturbing current solutions and applying local search techniques to improve solution quality.	ILS employs scoring functions that assess the fitness of candidate solutions based on the energy and stability of the docking complex, considering factors like steric clashes and intermolecular interactions.	[86, 87]
10.	Evolutionary Strategies (ES)	Rosetta and AutoDock	ES is a family of optimization algorithms inspired by biological evolution. It utilizes mutation, recombination, and selection to evolve a population of candidate solutions toward optimal solutions iteratively.	ES typically utilizes scoring functions similar to those used in other optimization algorithms, evaluating the fitness of candidate solutions based on the energy and stability of the docking complex.	[56, 84]

Table 1.

Outline of specific software applications in docking and their primary principles.

4.3 Post-docking analysis

4.3.1 Scoring functions

Scoring functions are crucial in molecular docking studies to evaluate the quality and feasibility of anticipated ligand-receptor interactions. These functions aim to characterize the binding affinity between a ligand and a receptor by evaluating various energy contributions [83]. A scoring function consists of terms related to van der Waals forces, electrostatic interactions, hydrogen bonding, dissolution effects, and sometimes additional terms for specific interactions. The docking algorithm’s various ligand binding postures are ranked and prioritized using the scoring function during post-docking analysis. Finding the most energetically advantageous binding modes is the aim. Scoring functions are estimates of the actual binding energy and might not adequately represent all facets of molecular interactions, and thus, it is essential to understand their limits. In order to direct the optimization process and find energetically advantageous docking configurations, scoring functions are essential. A brief overview of the scoring function employed by various docking techniques is provided in Table 2.

Software’s	Designer company and year of introduction	Docking approach/scoring function	Relevancy and novelity of softwares	References
AutoDock	D. S. Goodsell and A. J. Olson The Scripps Research Institute 2009	It is based on force field methods and follows the Lamarckian genetic algorithm used for scoring function.	Open-source software, improved accuracy and speed, and user-friendly interface	[70]
Glide	Schrodinger Inc. 2022	It is based on Monte Carlo sampling & GlideScore used for scoring function	Ligand-based design quantum mechanics molecular dynamics simulations	[72]
Gold	Cambridge Crystallographic Data Centre 1997	It is based on automated ligand docking program that uses a genetic algorithm for scoring function	Based on genetic algorithm & six different scoring functions	[88]
Mol Dock/Molegro	Molexus a Danish company 2005–2008	It is based on semi-empirical differential evaluation algorithm	High docking accuracy, easy-to-use interface, and proprietary scripting languages	[73]
Flex X	T. Lengaurer and M. Rarely BiosolvetT 2023	It is based on incremental construction & screen score, flex score PLP	Based on Monte Carlo algorithm & performed templet docking covalent docking (structure-based docking)	[89]
Surflex-Dock/SYBYL software 1990	BioPharmic	Hammerhead docking system having empirical scoring function	Show different potential binding modes, advanced feature	[90]
ICM-Pro	Molsoft LLCI 1985	Emperical	Monte Carlo algorithm, with crystallographic analytical tools	[71, 75]

Table 2.

Summarizes some commonly used protein-ligand docking tools.

4.3.2 Visualization tools

In molecular docking studies, visualization methods are essential for analyzing and understanding the complicated three-dimensional structures of ligand-receptor complexes. These tools allow scientists to gain additional insight into molecular interactions’ general geometry, interactions, and binding mechanisms. Examples of frequently used visualization tools are Pymol, UCSF Chimera, JMOL, and RASMOL. While it is outside the scope of this chapter to cover all the tools, several excellent references provide information on visualization software.

5. Challenges and limitations

Common challenges and pitfalls in molecular docking include:

Protein flexibility: ignoring protein flexibility can lead to inaccurate predictions as proteins can undergo conformational changes upon ligand binding. Incorporating protein flexibility in docking algorithms is crucial to account for induced fit effects.

Active site water molecules: handling active site water molecules can be challenging. Incorrect treatment of water molecules may introduce false energetics and affect the accuracy of docking results.

Scoring function selection: choosing an inappropriate scoring function can impact ligand ranking and pose selection. Using scoring functions suitable for the specific docking task and target is essential.

Ligand conformational sampling: inadequate sampling of ligand conformations can result in missing potential binding modes. Proper ligand conformational sampling techniques are necessary to explore the full conformational space of ligands.

Structural data quality: reliance on low-quality or incomplete structural data for proteins and ligands can lead to unreliable docking results. Using high-quality structural data from reputable sources like the PDB is crucial for accurate predictions.

Algorithm selection: choosing the wrong docking algorithm for a specific target or ligand can result in suboptimal docking performance. It is important to consider the characteristics of the target and ligands when selecting a docking method.

Overlooking protein-ligand interactions: failing to consider all relevant protein-ligand interactions, such as hydrogen bonding, hydrophobic interactions, and electrostatic interactions, can lead to inaccurate binding predictions.

Scoring function limitations: scoring functions may have limitations in accurately predicting binding affinities or distinguishing between active and inactive ligands. Understanding the strengths and weaknesses of different scoring functions is important for interpreting docking results.

6. Case studies

As invaluable insights into using computer methods for comprehending protein-ligand interactions and drug discovery in molecular docking. A handful of the research that is presented below demonstrates the molecular docking and virtual screening procedures that are used to look into viral proteins, create new vaccines, and assess possible treatments. These techniques provide crucial possibilities for addressing major global health issues.

6.1 Case study 1: chikungunya virus E glycoprotein

Chikungunya virus (CHIKV) outbreaks have been widespread across Asia, Africa, and Central and South America, causing significant public health concerns. Due to its positive sense RNA nature, CHIKV’s replication within the host has led to mutations in its genome, posing challenges in vaccine development and treatment strategies. While vector control remains effective in curbing CHIKV transmission, eradicating the vector species presents formidable obstacles [91]. Numerous conserved epitopes across different CHIKV strains have been identified through extensive investigation. Leveraging these putative antigenic epitopes holds promise for developing innovative vaccines to combat and eradicate CHIKV infections.

6.2 Case study 2: dissection of capsid protein HPV

Human papillomavirus type 52 (HPV 52) stands as one of the most concerning strains contributing to cervical cancer globally. This study explored potential vaccination approaches against cervical cancer utilizing computational techniques such as immuno-informatics and molecular docking [92]. The research presents a robust analysis and practical methodology for efficiently generating vaccines.

6.3 Case study 3: in silico rational design of a novel tetra-epitope tetanus vaccine

Exposure to various epitopes within the tetanus toxoid vaccine can produce undesirable antibodies by lymphocyte clones. Ideally, tetanus vaccination should cater to individuals with specific human leukocyte antigen alleles. Clostridium tetani bacteria, responsible for tetanus, release toxins causing severe muscular contractions, often termed as “lockjaw.” This study explores the rational design of a tetra-epitope tetanus vaccine, aiming to provide comprehensive protection against tetanus while considering genetic diversity [93].

6.4 Case study 4: evaluation of potential anti-hepatitis A virus

Hepatitis A virus (HAV) remains a common cause of acute hepatitis globally, occasionally leading to acute liver failure. Despite the availability of the HAV vaccine, its prevalence in certain regions persists. This research assesses a potent HAV 3C protease inhibitor effective against human hepatoma cells infected with HAV genotype IIIA HA11-1299. Highlighting the significance of hydroxyl groups in inhibiting protease activity, the study suggests small molecule inhibitors as a viable strategy for developing novel anti-HAV therapeutics [94].

6.5 Case study 5: monkeypox virus

Monkeypox, caused by a double-stranded DNA virus belonging to the Poxviridae family, shares structural similarities with the smallpox virus. Recent outbreaks of monkeypox in African nations pose a significant public health risk. Molecular docking studies revealed hydrophobic interactions between FDA-approved medications and the proteinase amino acid residues of the monkeypox virus. This research sheds light on potential therapeutic targets for managing monkeypox infections in public health emergencies [95].

7. Conclusions

In conclusion, the chapter encapsulates the foundational aspects and significant applications of molecular docking in contemporary research. Beginning with elucidating the fundamental principles underlying protein binding and activation, the chapter navigates through the protein-protein interactions and ligand binding mechanisms essential for drug development. By exploring the diverse array of software tools available for molecular docking, the chapter sheds light on the underlying algorithms that power these platforms. It underscores the critical role of molecular docking in accelerating drug discovery processes, facilitating computer-aided drug design, and enabling virtual screening methodologies. Furthermore, through compelling case studies, the chapter vividly illustrates the indispensable role of molecular docking in advancing medical development. It underscores its role as a transformative tool driving innovation in medicine development and its significance in shaping the landscape of modern research paradigms.

Acknowledgments

The authors would like to thank Marwadi University Rajkot, Gujarat, India, for providing the necessary assistance to accomplish this study.

Conflict of interest

The authors declare no conflict of interest.