INTRODUCTION:

Molecular docking, does mean a crucial computational method in drug design and discovery, which is predicting molecular orientation temperature due to small molecules interacting with target proteins. It identifies potential drug molecules, optimizes binding interactions, and makes a significant difference in new therapeutic agent development.

That is why this technique is essential for the interaction understanding between drug molecules and target proteins, making it possible to discover potential binding sites, optimize affinity, and improve specificity. Although molecular docking is a relatively faster and economical screening of drug candidates, it has been supplemented by manual drug studies, which generally provide more information about the particular drug-target interactions, allowing greater precision¹. Both molecular docking and manual drug studies have their specific merits and demerits. They are often paired together to get the best of both worlds and thus enhance effectiveness in drug discovery. The choice between procedure between molecular docking and manual drug studies is dependent on the type of question asked and other factors like available resources and the kind of complexity of drug-target interactions in question. By understanding the complementary nature of these approaches and utilizing them strategically, researchers can optimize the drug discovery procedure and accelerate the development of novel therapeutic solutions².

Docking studies are crucial for understanding biological and pharmacological interactions. Efforts have been made to enhance algorithms for docking prediction, which anticipates the preferred orientation of molecules in a stable complex. Scoring functions can estimate binding affinity strength based on these orientations³. Searching algorithms are generating possible postures, ranked based on some scoring functions. There are many programs which were developed during the last decades, some of which include the most well-known names as auto dock, auto dock vina, etc. Docking calculations involve obtaining the targeted protein structure, which typically consists of a macro molecule and micro molecules. The Protein Data Bank (PDB) provides 3D atomic organizes for these molecules. However, the experimental 3D structure of the target is not available. Computational prediction methods like comparative and ab initio modelling can help⁴.

History:

In the late 20^th century, the drug discovery era of molecular docking is the origin to find out various target molecules for the certain structure of the drug candidate. Over the years, significant advancements and key milestones have marked its evolution. In the 1980s, molecular docking algorithms were developed to predict binding modes of micro molecules to target proteins, aiding drug design. Advancements in computational control and algorithms in the 1990s improved the accuracy and efficiency of predicting ligand protein interactions⁵. Molecular docking became popular as a strategy for drug discovery at the end of the last century as scientists began accessing technologies like crystal structures. It has been much adopted in experiments by researchers and pharmaceutical companies for drug development activities, strengthening simulation of the binding process and including various potential drug candidates into programme development⁶. This review discusses molecular docking methodologies, approaches, algorithms, types, significance, importance of docking and applications in forming a drug–protein complex for various drug discoveries.

THEORY OF MOLECULAR DOCKING:

Molecular docking, a computation method, is, as mentioned, one of the most common used in the Structure Based Drug Design for predicting the ligand-receptor complex structure. Molecules are bonded in a cell to form a sustainable complex to allocate in optimal interactions with receptors.⁷.

Docking encompasses two key processes: sampling ligand conformations in the active site of the protein and ranking them using a scoring function with sampling algorithms reproducing the experimental binding modes and the top ranking among the generated conformations⁸.

SEARCHING ALGORITHMS:

The primary target in developing a docking algorithm is to have a fast method for the discovery of novel lead compound or conformation with as much accuracy as possible. These are definite by a set of instructions and parameters². In relations to the flexibility of the receptor and/or the ligand, concerning their movements, docking algorithms can be considered in two large sets: rigid-body and flexible docking which are grounded on the different types of algorithms².

Rigid body docking focuses on vital geometric complementarities without considering the flexibility of the ligand or receptor which limits the specificity and accuracy of results but can identify the ligand binding sites close to crystallographic structures⁹. "The Root Mean Square Deviation (RMSD) is working to compare docking simulation results with crystallographic structures. This method is commonly used for rapid initial screening of small molecule databases. For example, by using the molecule surface of the receptor, a negative image of the binding pocket is formed by overlapping spheres of different radii that touch the molecular surface at two points. To correlate matching sets, atoms of ligand are aligned to the centers of these spheres ensuring all inter-sphere center distances correspond to those between ligand atoms.

There are many ways of assessing the initial screening. The first method is just rigid-body docking, and flexible docking is for improvement and lead optimization. More computational power is needed for this process, and different conformations of ligand, receptor, or both simultaneously are being considered for this specific improvement. A flexible docking takes an additional step than that of rigid-body docking in that it considers rotational degrees of freedom and all translations and not only for the ligand but also for the receptor¹⁰. Common procedures for searching conformational space include

I. Systematics Search algorithm

II. Random or Stochastic algorithm

III. Simulation algorithm

I. Systematic search algorithm:

Systemic search is characterized by small changes in the structural parameters, slowly it changes the conformation of the ligand to explore all the degrees of freedom determined by the amount of rotation, angles and dimensions in the molecule. It is of two types:

· Exhaustive search algorithms

· Fragmentation based algorithms

Exhaustive search algorithms:

The method identifies ligand conformations by rotating rotatable bonds and identifying ligand positions using scoring approaches. Large conformational spaces prevent exhaustive systemic searches, so high-resolution docking search is applied to reduce conformational space. Heuristics, used by algorithms like GLIDE, focus on high-scoring ligand positions in specific areas¹¹.

Fragment based algorithms:

Fragment based algorithms is further divided into three types such as

· Incremental Construction

· Distance Geometry (DG)

· Fast shape matching (SM)

Incremental construction:

They are making the construction of ligand structures from fragments, which are derived from the fragmentation of the ligand of interest. They are docked into the binding site whether one by one and increased or by docking all fragments together and covalently joining them to yield ligand conformations. They are added within geometric restrictions to the binding sites concerning their steric complementarities and their binding affinities. There are two primary techniques for the incremental push-search algorithm:

De novo ligand design: Different molecular fragments are docked into the active region and covalently linked. Core and side chain docking: The ligand is divided into rigid (core) and flexible portions (side chains). Dock core first into active site, then add flexible parts incrementally.

Distance Geometry (DG):

During the evolution of DG, inter and intra molecular distances are utilized. It employs a smaller subset of distance constraints and is agreeable with a large number of constrictions by adding or deriving bounds from specified bound constraints. Distance geometry serves an FLOG and creates database conformations to work with it like DOCK¹².

Fast Shape Matching (SM):

This algorithm is dependent on the geometric overlaps that originate from the molecular surfaces of two molecules. The ligand complementarity predicting algorithm thus proposes a number of potential conformations of the binding site. For instance, ZDOCK uses Fast Fourier Transform algorithms along with a geometric surface-based model for shape complementarity, desolvation, and electrostatics parameters¹³.

II. Stochastic or Random search methods:

These algorithms are used to generate various molecular conformations, that involves making random modifications to a single ligand or a group of ligands, which are then evaluated using a predefined probability function hence changes are accepted¹⁴. These are the methods utilise different probability criteria for accepting changes.

They are – Genetic Algorithms

Genetic Algorithms

Monte Carlo Simulation

Tabu Search method

Genetic Algorithms:

They are based on residents and biological evolution principles. It specifies the layout of protein and ligand by parameters and state variables which describe translation, rotation, and conformation and those parameters are defined in a chromosome and assessed by fitness function. Hence the total particles of proteins and ligands known as fitness value in molecular docking. The new chromosomes are not necessarily derived from the original chromosomes; instead, random pairs of chromosomes are utilized to produce a new chromosome or offspring by crossover. The new chromosome inherits genes from their parents. Mutation will be accepted only if improves the fitness value, wherein some offspring undergo random mutations; that is, one gene is altered by an arbitrary amount. Thus, more fit solutions reproduce while less fit solutions get eliminated, just like in nature mimicking evolution¹⁵. It comprises many components in its scoring function, including mutation rates, crossover rates, and rounds for evolution. In the GOLD docking program, it also requires input size and approximate position where the receptor's active site is located. The Lamarckian Genetics Algorithm used in the AUTODOCK switches from genotypic to phenotypic and back again¹⁶. As pertains to energy function, gamma and crossover procedures are primarily seen in genotypic space while the enhancement occurred in phenotypic space.

Monte Carlo simulation:

The process starts with a random ligand configuration in the active site, evaluated based on properties like energy, and then applies random motion like conformation, translation, and rotation, generating a new configuration¹¹. This new system is once again scored; it will be accepted or discarded on the basis of Metropolis criterion, which uses a probability function based on Boltzmann. This process goes on until a desired number of configurations are reached. These procedures are incorporated as early minimization step in the course of molecular dynamics agendas such as GROMACS and GROMOS. Monte Carlo is also employed as flexible docking algorithms in programs like MCDOCK and ICM. It guarantees exact and very accurate results in thermodynamic conditions.

Tabu Search Algorithm:

This is a heuristic meta-method for hard optimization problem solving developed by Glover. Heuristics or approximate methods are employed for difficult machinery problems. It's little random modifications on the current ligand conformation and fitness function ranks them. The previous rejected conformations are known as tabu. Starting with a possible solution and iteratively refining it through local modification using local search techniques. These methods, however, usually reach a local minimum. But it is accepted in case of tabu if even its value were less than for any previously accepted change if it showed best improvement. Or else, the best non-tabu modification is accepted and recorded. This cycle repeats till the real minimum is established.

III. Simulation methods:

It takes place with the initial topological state occurring in the first cycle as the change takes place. The fundamental algorithm for simulation is to measure its execution time as an instruction within a trace, while the simulated annealing search process simulates a biomolecular system through specific molecular dynamics simulation. Results may be rejected based on the evaluation principles or based on calculations for steric clashes and distance matching. It has a range of applications spanning conformational analysis, prediction of protein structures, and also molecular docking search methods. On comparison with Monte Carlo methods, simulated annealing takes into account the configuration and flexibility of both the ligand and protein which would make it more accurate¹¹.Top of Form

It includes- Molecular Dynamics Simulation

Simulation Annealing Bottom of Form

Molecular Dynamics Simulation:

The mechanism, based on Newtonian laws, determines the trajectory of the system. Atomic positions are subsequently determined for a bunch of small-time steps through an application of Newton's second law of motion by the use of these molecular forces and corresponding masses of atoms¹⁷. Much of the area of molecular dynamics (MD) simulation is a difficult task for crossing high-energy barricades within the timeframe of simulation, resulting in local minima positions. One way to improve this outcome is to marry MD with other methods such as simulated annealing. An MD protocol created by Mangoni et al. in 1999 was intended to fit small and flexible ligands to flexible targets in aqueous media. The movement of the center of mass of the ligand, replotted separately from any internal and/or rotational motions, was then coupled with different temperature baths. The mechanism of the so-called relaxed-complex approach, which considers binding conformations that are quite rare in unbound target proteins, involves carrying out an MD simulation of the target without ligand for 2 ns and thereafter docking the ligand. This yielded the first clinically approved HIV integrase inhibitor, Raltegravir. Long MD simulations study drug-binding events at protein targets¹⁰.

Simulation Annealing:

The simulated annealing search procedure is a molecular dynamics simulation for conformational analysis, protein structure prediction, and molecular docking searches. It consists of a gradual temperature drop for each docked conformer, after which solutions that do not meet with the scoring criteria, conflict sterically with another feature, or mismatch some distances are rejected. Generally, this technique provides better results than the Monte Carlo method but is rather time-consuming. This can be further combined with Monte Carlo (Auto Dock) for updating the ligand orientation by random changes in every annealing temperature cycle¹¹.

SCORING FUNCTIONS: Bottom of Form

Scoring functions are quantitative procedures which estimate the energy of the interaction of the molecules after the docking process. It is routinely used in structure based virtual screening, because it allows for the evaluation and ranking of predicted conformations of ligands. Scoring functions are easy and time efficient to compute despite the fact that they are not very accurate; they can be used to generate consensus scores. They distinguish between the correct native solutions with RMSD of 2 Å within the crystal complex and others in a reasonable computation time. Ranking or scoring methods are initially employed in virtual screening exercise and are employed in activities that estimate free energy of protein ligand docking complexes such that enthalpy together with entropy in ligand binding procession is considered. The methods of perturbation by free energy in an additive equation of the many components of binding are binding models that are additive¹⁰. A complete equation represents

∆G_bind= ∆G_solvent+∆G_conf+ ∆G_int+∆G_rot+∆G_trans/rot+ ∆G_vib

Where,

∆G_solvent = Interaction with solvent, additionally, with protein and ligand itself.

ΔG_conf = describes the outcome caused by variations in the ligand or protein conformation.

ΔG_int = provides the energy of free interaction particular to the protein-ligand binding

ΔG_rot = equivalent to a measure of lost free energy due to freezing rotatable bonds when it comes to entropic contributions.

ΔG_trans/rot = translational loss and rotational free energy by way of the interaction of two bodies (protein and ligand) into one entity, complex.

ΔG_vib = free energy contribution due to vibrational mode changes taking place.

Scoring Functions Serve Three Basic Applications in Molecular Docking. They Are:

1. To determine the ligand's binding mode as well as its site on to the protein.

2. Premises an total binding affinity of the ligand-protein pair, which is especially useful for lead optimization.

3. Search among a large ligand database for possible drug hits or leads for a given target protein, that is virtual database screening¹⁸.

Types of scoring functions:

The scoring functions are classified into five types

1. Force-field based scoring

2. Empirical scoring

3. Knowledge based scoring

4. Machine learning based scoring

5. Consensus scoring

1. Force Field-based Scoring Function:

Force-field scoring functions involve physical atomic interaction, electronic field effects, bond stretching/bending, and torsional forces. Molecular mechanical force fields output potential energy of receptor ligands or incidental energy experienced in association with inside energy due to steric offshoot needed for receptor-ligand building.

S = E_PL+ E_L

The protein-ligand interface involves two electrostatic terms, EPL and EL, evaluated through a columbic model. The dielectric function is distance, with van der Waals terms tuned based on LJ potential function. Force-field scoring functions have low computational speed, so non-bonded interactions are converted into common cut-off distance. Extensions include hydrogen-bonding, solvation, and entropy for high ligand binding specificity¹². The solvents' effects are best treated through explicit consideration of water molecules like Free Energy Perturbation (FEP) and thermodynamics integration (TI). On the other hand, implied solvent continuum representations like Poisson-Boltzmann (PB) or Generalized Born (GB) can be used in scoring molecular docking Permits²⁰.Common software programs for docking functions include AUTO DOCK, DOCK, G SCORE, GOLD SCORE, and D SCORE, each with unique approaches to hydrogen bonding and energy types. Other techniques like Linear Interaction Energy and Free-Energy Perturbation can enhance predicted binding energies¹⁰.

2. Empirical Scoring Function:

Empirical scoring functions, introduced by Bohm in 1992, provide experimental results on binding energies or conformations. These functions consist of several parameterized terms, including hydrogen bond distinctions, contributions based on hydrophobic molecular SAs size, non-enthalpic terms like the rotor term, and solution and dissolution from a continuum model. These functions aim to provide a comprehensive understanding of binding processes and their underlying enthalpy²¹. The empirical scoring function approximates the stability of the complex formed between a protein and a ligand based on several individual numeric terms added together. The functional form of empirical scoring functions has one less layer of complexity to interpret than forcefield based scoring functions. Many of the contributory terms may be similar to or derived from forcefield molecular mechanic terms. In this, binding energy decay the complete binding free energy into various energetic ways as:

S = ∑_iW_iE_i

Empirical scoring functions are used to evaluate binding of protein-ligand in complexes. These functions are based on regression analysis and are used to determine the binding affinity of ligands not used in the training set. The basic forms of these scoring functions are consistent, but the energetic terms have different realizations. Software packages LUDI, CHEMSCORE, F-SCORE, and GLIDESCORE are used to treat terms relative to each other. The force-field-based scoring function uses item-theoretic under-depiction, while the empirical scoring function relies on remote finding of the functional form for fitting. However, empirical scoring functions have major drawbacks, such as reliance on experimental dataset regression analysis and fitting methods, making it difficult to establish a rule for fitting all components needed for protein-ligand binding weighting. There is no assurance of how well an empirical scoring function can approximate the affinity to bind with the ligands not used in the training set¹⁹.

3. Knowledge based scoring function:

· The experimental geometric structures will be reproduced, while binding energy will not.

· The free energies of such molecular interactions can be calculated from structure data on the Protein-ligand complexes.

· Interatomic distance distribution is translated into energy functions via inversion of the Boltzmann law.

· Drug score adds solvent approachability rectifications to the atom pair abilities.

Unfortunately, the main drawback in this score arises because it is derived solely from limited information existing from the small crowds of protein ligand complex structures²¹.

These are used for protein construction prediction and protein-ligand research. These functions rely on pairwise potentials for protein-ligand complexes, which represent potential interactions. Pairwise potentials are resulting from the inverse Boltzmann statistics and are calculated by analysing the frequency of a particular atom pair in a database.

S = ∑ w(r)

W(r) = −k_BT In[g(r)] = −k_BT In [

w(r) indicates the specific pairwise potential between the ligand and the protein. kB is Boltzmann's constant, T is the absolute temperature of the system under study, f(r) is the occurrence frequency of the atom pair at distance r, and f*(r) indicates the average occurrence frequency in the corresponding state of the reference system.

In general, score functions based on knowledge are faster than both empirical and force field-based score functions and still reasonably accurate. This is because they derive their potentials directly from structure relatively trying to replicate known affinities through fitting. Thus, this scoring functions have fairly low dependencies on the training set, owing to their origin from a very large, structurally diverse data set²³.

Known based scoring function is a phenomenon that involves the use of a number of software programmes. Like: PMF, SMoG, DFIRE, BLEEP, ITScore, MScore, DrugScore, KScore, GOLD/ASP and so on.

4. Machine learning based scoring function:

This scoring functions, introduced in 2004, are a relatively new approach based on the Quantitative Structure-Activity Relationship (QSAR) study of thin interaction between ligand and protein investigations. These methods estimate statistical models for ligand-protein binding scores, encoding the characteristics of protein and ligand and interactions with specific descriptors. Variable selection methods include neural networks, Bayesian classifiers, random forests, and support vector machines. The approach requires a set of ligand- protein complexes and binding information, such as NNScore, RF-Score, SFCscoreRF24, and ID-Score. However, Gabel et al. found that this scoring functions are low on docking tests, indicating that the native pose among docking decoys is not discerned by conventional scoring functions. This could be due to ML-based functions being over-trained on interaction-independent atom-pairs counts, which do not consider the docking pose. Systemic improvement and multiple validation are needed to address this issue²⁵.

5. Consensus Scoring functions:

Consensus scoring is a new method introduced by Charifson et al in 1999, which aims to reduce errors in singular scoring and increase the chance of recognizing true modes/binders/multi-copy binders. It involves combining individual scores into a consensus score strategy to distinguish 'true' modes/binders from majority having different molecules. Common consensus scoring methods include MultiScore, Chemscore, GFscore, Xcscore, Gold-like, and FlexX. Studies have shown that this scoring can progress presentation by nullifying shortfalls observed in single scoring and reducing false-positive outcomes detected with a single scoring function. However, authors argue that consensus scoring may only yield a small benefit if predictions of distinct scoring functions align²⁰.

MODELS OF MOLECULAR DOCKING:

The Molecular docking is designed based on some models. The models are

· Lock and key model

· Induced fit model

· The conformation ensemble model

Lock and Key model:

Fig 2. The lock and key model

Lock and Key model is a type of model applied to determine interaction between an enzyme and a substrate. Key point of Emil Fischer I accept the Lock and Key model where the binding between two substances that took its place in 1980 describes the biological processes. A substrate fits into the active site of a macromolecule as a key fit into a lock. The substate that binds into the enzyme’s active site to produce an enzyme-substrate complex. Biological keys are physically endowed with certain special stereochemical features which determine the function of the machines. The early docking methodologies that were related to this theory, along with the ligand and receptor, were considered rigid.

Induced fit model:

Fig 3. Induced fit model

Induced fit model A model of enzyme-substrate interaction in which the enzyme and substrate undergo other structural changes to ensure a close fit. In 1958, Daniel Koshland took the lock and key theory a step forward has arrived at the "Induced fit theory", which describes that the active location of the target healthy protein is slowly re-shaped by small conformational changes in addition to the ligands as the ligand binds to the target protein, the fit will stabilization. Thus, it more accurately depicts the binding events than does the rigid treatment⁷.

The conformation ensemble model:

Conformational ensembles are computational models also referred to as structural ensembles since they describe the structure of intrinsically unstructured proteins. These proteins are intrinsically flexible and it refers to the structure of those proteins which does not possess a stable tertiary structure. The proteins have also been found to display much larger conformational changes²². Buyong Ma et al. in 2003 mentioned the "Conformational ensemble model". It thus supposes that the protein is composed of an ensemble of already existing conformational states, which enables the flexibility of protein to switch states.

MOLECULAR DOCKING APPROACHES:

This is in fact what is termed a "molecular docking approach." It has several kinds, with its own distinct methods and applications:

1. Monte Carlo approach:

It starts by generating a random configuration, translation, and revolution of the ligand at the active site. The movement in this approach starts with some score from the evaluation of the current configuration and iteratively creates and evaluates a new configuration. To decide whether or not to keep the new configuration, we can invoke the Metropolis Criterion²³.

2. Metropolis Criterion:

New approaches with higher scores are accepted immediately, while those not novel use Boltzmann's law for probabilistic acceptance. Rejected arrangements fail the likelihood function test, but those that pass pass, making the arrangement acceptable. If the score is higher, the arrangement is accepted²⁴.

3. Matching approach:

Main goal is making this method complementary; place Ligand atom at optimal site position, generate ligand receptor configuration, and the refine.

4. Ligand fit approach:

according to shape complementarity, ligand fit has been an umbrella word under which rapid and accurate methodologies for docking small particles ligand into protein active spots are grouped.

5. Point complimentarily approach:

The techniques are focused on to assess a shape and/or chemical complimentarily between interact molecules²².

6. Blind Docking:

Blind docking is the name given to docking carried out without the assumption of a binding site. This was established to show potential binding sites of peptide ligand and mechanisms of action and to fully scan the interface of target molecules.

7. Fragment - based Method:

Fragment-based approaches involve solvating ligands in photons or particles, attaching fragments, and finally linking fragments²³.

8. Distance Geometry:

For instance, structural information can be expressed as intra- or intermolecular distances. As a DG framework, it allows the construction of such distances and calculation of the 3D models compatible with them²².

9. Inverse Docking:

Computer-aided methodology helps determine protein targets for small molecules, facilitating the assessment of potential toxicities and side effects. This knowledge, combined with a proteomic pharmacokinetic profile, can be used for docking studies with specific ligands, enhancing drug candidate assessment²⁴.

TYPES OF DOCKING:

1. Rigid Docking

2. Flexible Docking

1. Rigid Docking:

Rigid docking is a computational method for molecular modelling that predicts atomic resolution binding modes and affinities between ligands and receptors. It assumes rigid structures through the docking development, ignoring protein configurational variations, making it the fastest but most likely to ignore these changes²⁵. The goal is to rearrange one compound in 3D space to match other compounds in a scoring system²⁶.

Principles of Rigid docking:

The docking procedure relies on the ligand fitting to the receptor's likely position, ensuring binding. This involves examining the ligand's configurational space and identifying the most suitable conformation for acceptable interactions like hydrogen bonding, van der Waals, hydrophobic interactions, and avoiding steric clashes²⁸.

Method of Rigid docking:

Rigid docking involves making ligand and receptor structures by removal of water molecules, addition of hydrogen atoms, conveying partial charges, and enhancing geometry. Docking algorithms use search algorithms to discover ligand conformational space and find the optimal binding posture. A scoring function evaluates and ranks candidate ligand poses based on predicted binding affinity. Post-processing techniques fine-tune predicted binding poses, study intermolecular interactions, and use visualization tools for complex examination and analysis of main binding remains and interactions²⁹.

Limitations of Rigid docking:

It is a crucial aspect of molecular modelling, treating model ligands and receptor sites as rigid bodies. This approach may result in less accurate predictions due to the lack of changes with ligand binding. The accurateness of rigid docking prediction be contingent on the scoring function used, which may not accurately handle all molecular interactions, causing false alarms or misses. Docking simulation is time-consuming and requires efficient algorithms and parallel computational resources²⁶.

2. Flexible docking:

Flexible docking is a computational system used for molecular modelling applications, predicting binding modes between a receptor and ligand. It captures conformational changes in receptor and ligand constructions, enabling better predictions of binding modes. Unlike rigid docking, flexible docking allows for conformational changes during binding. The ability of molecular flexibility is evaluated by adding transformations and determining confirmation for receptor and ligand molecules in a complex²⁷.

Principles of Flexible docking:

Flexible docking is a dynamic approach that predicts binding posture and potential changes in conformational for receptor and ligand, permitting for a more realistic representation of their interaction vigor.

Methods of Flexible docking:

Flexible docking is a method that uses conformational sampling and classical search techniques to predict the best binding pose of a ligand within a flexible receptor binding site. It considers geometric complementarity, electrostatics, conformational strain, Van der Waals, solvation. Induced-fit modeling captures the most accurate conformational change of the receptor during binding. Post-processing methods refine predictions and analyze intermolecular interactions.

Limitations of Flexible docking:

Flexible docking is a method for studying docking between molecules, but it has limitations. High computational costs in large biomolecular systems and extensive conformational sampling make it less scalable. Current scoring functions can't describe different binding molecular interactions, resulting in errors in predict affinity of binding and posture. Sampling of conformation is crucial for estimating diverse ligand and receptor conformations, but it's difficult for large biomolecular systems due to complex energy landscapes²⁸.

There are some other types of molecular docking which are also involved with the approaches are:

1. Induced fit docking

2. Ligand based docking

3. Protein-protein docking

4. Blind docking

1. Induced fit docking:

Induced fit docking is a method that uses a combination of rigid and flexible docking methods to predict binding modes and affinities of a ligand with a receptor. It involves initial rigid docking followed by flexible side chain or backbone movements, considering changes in conformations²⁹.

Principles of Induced fit docking:

It is a method that reveals the dynamic interactions between biomolecules for a bound complex, considering the movement of ligands and receptors simultaneously. The process involves pre-docking conformational sampling and docking with a flexible receptor, where the receptor domain model becomes flexible, allowing for optimized interactions with a stable bound complex.

Methods of Induced fit docking:

It is a method that collects family collections of conformations obtained through molecular dynamics simulation, analysis of normal mode, or conformational search algorithms. It combines the definition of optimal ligand binding pose with conformational sampling and classic search techniques. The scoring function assesses ligand compatibility and guides the search to the best binding poses. Post-processing methods improve prediction and analysis of intermolecular interactions, making it useful in biology³⁰.

Limitations of induced fit theory:

The imperfection of this theory leads to computational cost dispassion in predicted binding affinity and pose problems, limiting the scalability of docking methods. Current scoring functions fail to accurately predict interactions, and modelling receptor flexibility becomes more complex for larger biomolecular systems with complex energy landscapes³¹

2. Ligand based docking:

Ligand docking is a simulation-based molecular modelling technique that forecasts the binding manner and affinity of a molecule to its target enzyme receptor, especially in situations where the target receptor construction is not known or difficult to obtain³².

Principles of Ligand based docking:

It uses pharmacophore mapping and molecular similarity to identify functional groups and structural features and crucial for binding and biological activity, predicting binding affinity and selectivity towards target receptors.

Method of Ligand-based docking:

It is a method that uses chemical databases to identify and classify ligands based on their activities against specific target receptors. This algorithm aids in model-building and similarity searches, ranking candidates based on similarities to known active compounds and predicted binding affinities. The predictive accuracy and reliability of these assays are validated through experimental tests³

Limitations of Ligand based docking:

It faces challenges due to the need for at least one reference ligand, limited structural information, and limited simulating complex interactions. This leads to high "composite" targets, particularly in allosteric modulation, which can cause false predictions. Existing scoring functions may not accurately represent the difficulty of interactions in ligand-receptor, resulting in false positive or negative results in virtual screening outputs³³.

3. Protein-protein docking:

Docking is a computational method used in molecular modelling to forecast protein 3D construction and interaction modes, crucial for protein-protein interactions in biological processes like signal transduction and immune response³⁴. It also aids in the comprehension of protein complexes and signalling pathways.

Principles of Protein-protein docking:

Predictive docking of proteins generates probable binding configurations without experimental ascertainment, exploring conformational space for possible binding modes and affinities, including complementary surfaces, hydrogen bonds, hydrophobic forces, and amino acid interactions.

Methods of Protein-protein docking:

Protein docking algorithms search protein conformational spaces for optimal binding poses. Scoring functions evaluate predicted binding affinity and stability, dividing poses into native-like and non-specific interaction categories. Post-processing and analysis refine predicted poses and analyse intermolecular interactions using visualization tools to identify key residues of interaction interfaces³.

Limitations of Protein-protein docking:

This method have limitations such as conditional sampling, inaccuracy in scoring, and difficulty representing protein flexibility, leading to mispredictions of binding affinities and potential false-positive or false-negative determinations of complex structures. Validating these predictions is crucial for substantiating assumptions and understanding their biological relevance³⁵.

4. Blind docking:

Blind docking is a molecular modelling strategy that predicts the mode of binding and strength of association between a ligand and a receptor without defining a specific binding site. It docks the ligand to the entire receptor surface to discover potential interactions, while receptive design predicts binding mode and affinity level without selecting a specific site.

Principles of Blind docking:

Blind docking is a method that scans receptor surfaces for possible binding sites and forecasts the binding mode and affinity of ligands-receptor, without requiring prior knowledge of the binding site's position or structure³.

Methods of Blind docking:

The strategy involves creating a protein surface grid, selecting ligand conformations, applying a blind docking algorithm, and evaluating possible binding poses. This refinement helps predict binding affinity and stability, allowing researchers to view docked complexes and determine binding sites. This method can be applied to different protein regions, such as solvent-exposed or structurally flexible areas, for better understanding protein interactions³⁶.

Limitations of Blind docking:

Blind docking methods have disadvantages like high computational costs, sampling limitations, and lack of scoring functions. They are limited in scalability and need effective algorithms with equivalent computing capitals. Incomplete sampling of inaccessible binding sites or cavities can lead to pose failure. Current scoring functions don't capture the complete interaction of proteins and ligands, resulting in imprecise binding affinity prediction³. Blind docking is challenging to study flexible or dynamic binding sites as it doesn't involve conformation changes or bending between bound states⁴³.

Table 1: Basic characteristics for current protein-ligand docking tools

Entry	Program Ref.	Designer/company	Licence terms	Supported platforms	Docking approach	Scoring function
1	Auto dock [5]	D.S. Good sell and A.J. Olson The Scripps Research Institute	Free for Academic use	Unix, Mac OSX, Linux, SGI	Genetic algorithm Lamarckian genetic algorithm	Auto Dock (force-field methods)
2	DOCK [6]	I. Kuntz University of California San Francisc	Free for academic use	Unix, Linux, Sun, IBM AIX, Mac OSX, Windows	Shape fitting (sphere sets)	Chen Score, GB/SA solvation scoring, other
3	Flex X [7]	T. Lenganer and M. Rarey	Commercial Free evaluation	Unix, Linux, SGI, Sun	Incremental Construction	FleXScore, PLP, Screen score, Drug Score
4	FRED [8]	Open Eye Scientific software	Free for academic use	Unix, Linux, SGI, Mac, OSX, IBM, AIX, Mac OSX, Windows	Shape fitting (Gaussian)	Screen Score, PLP, Gaussian Shape score, user defined
5	Glide [9]	Schrodinger inc.	Commercial	Unix, Linux, SGI, IBM, AIX	Monte Cario Sampling	Glide Score, Glide Comp6
6	GOLD [9]	Cambridge Cryptallographic Data Centre	Commercial Free evaluation (2 months)	Linux, SGI, Sun, IBM, Windows	Genetic Algorithm	Gold Score, Chem Score User defined
7	Ligand Fit [11]	Accelrys Inc.	Commercial	Linux, SGI, IBM, AIX	Monte Cario Sampling	Lig, Score, PLP, PMF

SOFTWARES USED IN MOLECULAR DOCKING:

There are several software programs which have been used for molecular docking: Auto dock vina, Auto dock, Glide, FlexX, Dock, FRED, etc. and other modern docking tools like ICM, Pro dock, QXP, Slide, Surflex. Among the software being explained here.

APPLICATIONS OF MOLECULAR DOCKING:

Molecular docking is crucial for demonstrating biochemical reactions' possibilities and has various applications, such as guiding experimental investigations and identifying potential interactions.

Lead optimization:

The optimized orientation of ligand on its target can be predicted through molecular docking. Not only will this predict the different binding modes of ligand in the groove of target molecule, but also help in developing more potent, discerning and efficient drug applicants.

Hit identification:

Molecular docking in conjunction with scoring function can screen immense databases for in silico identification of drug-like molecules that would bind with the molecule of interest.

Drug-DNA Interactions.

Preliminary forecast of the drug's binding properties to nucleic acid is of considerable importance, but most anticancer chemotherapeutic agents target the nucleic acid and auxiliary processes as their primary cellular target³⁷.

· Bioremediation also predicts the kinds of pollutants that bring degradation of the produced enzymes.

· Docking is particularly used in the study of protein-protein docking.

· It is used to verify the side effects when co-administered with another molecule.

· It is used as drug design tools and as a study of geometry for a particular complex.

· Drug-DNA interaction establishes a correlation between the drug's molecular structure and its cytotoxicity.

· Molecular docking is usually done to prove the possibility of any biochemical reaction before it would ever be the subject of experimental part investigation³⁸.

Receptor preparation:

The docking software determines the procedure. A good option would be Selection of structure and binding sites. Very often, facility hydrogen is added; it varies from being very position-sensitive to less so.

Ligand Preparation:

Predicting pKa values for charged atomic species is possible using programs for various pH ranges, including quantum mechanical forcefield reduction, chemical transformations, and charge changes in the chemical structure⁷.

Case Studies on Molecular Docking:

Discovery of a novel HIV-1 Protease Inhibitor: Molecular docking was used to identify potential inhibitors of the HIV-1 protease enzyme, leading to the discovery of new drug. The resulting compound showed excellent antiviral activity in cell-based assays.

Identification of a new class of Anticancer Agents: Molecular docking revealed compounds with high affinity for BRCA2 proteins, showing significant growth inhibition, including those from Hippophae rhamnoides and Hippophae salicifolia, as potential anticancer candidates.

Identification of new Anti-bacterial Agents: Molecular docking was used to identify the molecule which acts against the E. coli; MurG enzyme in which it shows the potent inhibitory activity which further shows the effective antibacterial activity.

Identification of New class of Anti-Viral Agents:

This study identified a new compound that showed the potent inhibitory activity against the influenza virus; neuraminidase enzyme which exhibits the antiviral activity on further evaluation. It showed the significant inhibition of viral replicant in cell-based assay.

Neurodegenerative Diseases:

Molecular Docking used to recognize small molecules that bind to the beta – amyloid peptide, which is identified as a potential therapeutic strategy for Alzheimer’s disease.

Cardiovascular diseases:

Identified the new compounds that inhibit the angiotensin-converting enzyme (ACE), a key target for treating the cardiovascular diseases.

Pain Management:

Identified the new compounds that binds to mu-receptor, a potential therapeutic strategy for pain management.

Identification of Anti-Malarial Treatment:

Molecular Docking, identified the new compounds that inhibit the Plasmodium falciparum; dihydrofolate reductase enzyme, which acts as a key target for new anti-malarial agents.

Identification of Antiallergic drugs:

Molecular Docking, identified the new Zea m1 as allergen after analysing the binding mode of Zea m1 allergen (PDB ID 2HCZ) and human immunoglobulin E.

CONCLUSION:

Understanding the construction of compounds by two different molecules computationally is Molecular Docking-another form of computer-aided design. It classifies as rigid docking or flexible docking according to the flexibility of receptors. Search algorithms help locate information in data structures-logical instructions provided. Discovering a new lead molecule computationally is a very fast and accurate way of finding a new lead molecule. The docking molecules will be subsequent to scoring functions, which predict binding affinity. Enthalpy entropies play significant roles in ligand bindings and were involved in several free energy techniques in scoring functions. It also features the scoring earned by combining the various scores of diverse scoring functions, consensus which does not refer to a scoring function. The interaction between ligand and receptor is made on the basis of model and approach along with their method for preparation of protein structure, algorithm, scoring to be used. Some of the main software people use to predict the dock and also binding affinity of the molecule. Many are docked to create drug structures for treating many types of diseases, including Alzheimer's, cardio degenerative disorders, allergies, etc., as prevention against developing the disease.

CONFLICT OF INTEREST:

The author declares that there is no Conflict of interest.

ACKNOWLEDGEMENT:

The authors would like to express my sincere gratitude to Dr. Sri Ramachandra, and D. Chaitanya Dixit for their invaluable guidance and support to my manuscript.

REFERENCES:

1. Sahu MK, Nayak AK, Hailemeskel B, Eyupoglu OE. Exploring recent updates on molecular docking: Types, method, application, limitation and future prospects. International Journal of Pharmaceutical Research and Allied Sciences. 2024; 13(2-2024):24-40. doi.org/10.51847/Une9jqjUCl

2. Singh S, Baker QB, Singh DB. Molecular docking and molecular dynamics simulation. In Bioinformatics 2022 Jan 1 (pp. 291-304). Academic Press. doi.org/10.1016/B978-0-323-89775-4.00014-6

3. Shoichet BK, McGovern SL, Wei B, Irwin JJ. Lead discovery using molecular docking. Current Opinion in Chemical Biology. 2002 Aug 1; 6(4):439-46. doi.org/10.1016/S1367-5931(02)00339-3

4. Torres, P. H. M., Sodero, A. C. R., Jofily, P., and Silva-Jr, F. P. (2019). Key Topics in Molecular Docking for Drug Design. International journal of molecular sciences, 20(18), 4574. doi.org/10.3390/ijms20184574

5. Yu Y, Lu S, Gao Z, Zheng H, Ke G. Do deep learning models really outperform traditional approaches in molecular docking? arXiv preprint arXiv: 2302.07134. 2023 Feb 14.doi.org/10.48550/arXiv.2302.07134

6. Li T, Guo R, Zong Q, Ling G. Application of molecular docking in elaborating molecular mechanisms and interactions of supramolecular cyclodextrin. Carbohydrate Polymers. 2022 Jan 15; 276: 118644. doi.org/10.1016/j.carbpol.2021.118644

7. Raval K, Ganatra T. Basics, types and applications of molecular docking: A review. IP International Journal of Comprehensive and Advanced Pharmacology. 2022 Mar 28; 7(1):12-6. doi.org/10.18231/j.ijcaap.2022.003

8. Cheng C, Yu X. Research progress in Chinese herbal medicines for treatment of sepsis: pharmacological action, phytochemistry, and pharmacokinetics. International Journal of Molecular Sciences. 2021 Oct 14; 22(20):11078 doi.org/10.3390/ijms222011078

9. Oliveira JS, Pereira JH, Canduri F, Rodrigues NC, de Souza ON, de Azevedo Jr WF, Basso LA, Santos DS. Crystallographic and pre-steady-state kinetics studies on binding of NADH to wild-type and isoniazid-resistant enoyl-ACP (CoA) reductase enzymes from Mycobacterium tuberculosis. Journal of Molecular Biology. 2006 Jun 9; 359(3):646-66. doi.org/10.1016/j.jmb.2006.03.055

10. Yadava U. Search algorithms and scoring methods in protein-ligand docking. Endocrinol Int J. 2018 Nov 13; 6(6): 359-67. doi.org/ 10.15406/emij.2018.06.00212

11. Chemistry, KN Pharmaceutical, "Search algorithms in Molecular docking," www.youtube.com/KNP Pharmaceutical Chemistry, 2022.

12. Ewing, T.J., Makino, S., Skillman, A.G. et al. DOCK 4.0: Search strategies for automated molecular docking of flexible molecule databases. J Comput Aided Mol Des. 15, 411–428 (2001). doi.org/10.1023/A:1011115820450

13. Chen R, Li L, Weng Z. ZDOCK: an initial‐stage protein‐docking algorithm. Proteins: Structure, Function, and Bioinformatics. 2003 Jul 1; 52(1):80-7. doi.org/10.1002/prot.10389

14. Sousa SF, Fernandes PA, Ramos MJ. Protein–ligand docking: current status and future challenges. Proteins: structure, function, and Bioinformatics. 2006 Oct 1; 65(1): 15-26.doi.org/10.1002/prot.21082

15. Krovat EM, Steindl T, Langer T. Recent advances in docking and scoring. Current Computer-aided Drug Design. 2005 Jan 1; 1(1):93-102. doi.org/10.2174/1573409052952314

16. Brooijmans N, Kuntz ID. Molecular recognition and docking algorithms. Annual Review of Biophysics and Biomolecular Structure. 2003 Jun; 32(1):335-73. doi.org/10.1146/annurev.biophys.32.110601.142532

17. Weiner SJ, Kollman PA, Case DA, Singh UC, Ghio C, Alagona G, Profeta S, Weiner P. A new force field for molecular mechanical simulation of nucleic acids and proteins. Journal of the American Chemical Society. 1984 Feb; 106(3):765-84. doi.org/10.1021/ja00315a051

18. Guedes IA, Pereira FS, Dardenne LE. Empirical scoring functions for structure-based virtual screening: applications, critical aspects, and challenges. Frontiers in Pharmacology. 2018 Sep 24; 9: 1089. doi.org/10.3389/fphar.2018.01089

19. Yan Z, Wang J. Scoring functions of protein-ligand interactions. In Methods and Algorithms for Molecular Docking-Based Drug Design and Discovery 2016 (pp. 220-245). IGI Global. doi.org/10.4018/978-1-5225-0115-2.ch009

20. Kollman PA, Massova I, Reyes C, Kuhn B, Huo S, Chong L, Lee M, Lee T, Duan Y, Wang W, Donini O. Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Accounts of Chemical Research. 2000 Dec 19; 33(12):889-97.doi.org/10.1021/ar000033j

21. Singh, A. P., Ahmad, S., Raza, K., and Gautam, H. K. (2023). Computational screening and MM/GBSA-based MD simulation studies reveal the high binding potential of FDA-approved drugs against Cutibacterium acnes sialidase. Journal of Biomolecular Structure and Dynamics, 42(12), 6245–6255. doi.org/10.1080/07391102.2023.2242950

22. Vasant OK, Chandrakant MA, Chandrashekhar KV, Babasaheb GV, Dnyandev KM. A review on molecular docking. International Research Journal of Pure and Applied Chemistry. 2021 May 20; 22(3):60-8. doi.org/10.9734/irjpac/2021/v22i330396

23. Dias R, de Azevedo J, Walter F. Molecular docking algorithms. Current Drug Targets. 2008 Dec 1; 9(12):1040-7. doi.org/10.2174/138945008786949432

24. Gaba M, Gaba P, Singh S, Gupta GD. An overview on molecular docking. International Journal of Drug Development and Research. 2010 Feb; 2(2):219-31.

25. Sahu MK, Nayak AK, Hailemeskel B, Eyupoglu OE. Exploring Recent Updates on Molecular Docking: Types, Method, Application, Limitation and Future Prospects. Int J Pharm Res Allied Sci. 2024; 13(2):24-40. doi.org/10.51847/Une9jqjUCl

26. Huang N, Shoichet BK, Irwin JJ. Benchmarking sets for molecular docking. Journal of Medicinal Chemistry. 2006 Nov 16; 49(23):6789-801. doi.org/10.1021/jm0608356

27. Jin Z, Wei Z. Molecular simulation for food protein–ligand interactions: A comprehensive review on principles, current applications, and emerging trends. Comprehensive Reviews in Food Science and Food Safety. 2024 Jan; 23(1): e13280. doi.org/10.1111/1541-4337.13280

28. Ketata MA, Laue C, Mammadov R, Stark H, Wu M, Corso G, Marquet C, Barzilay R, Jaakkola TS. Diffdock-pp: Rigid protein-protein docking with diffusion models. arXiv preprint arXiv:2304.03889. 2023 Apr 8. doi.org/10.48550/arXiv.2304.03889

29. Yu Y, Xu S, He R, Liang G. Application of molecular simulation methods in food science: status and prospects. Journal of Agricultural and Food Chemistry. 2023 Jan 31; 71(6):2684-703. doi.org/10.1021/acs.jafc.2c06789

30. Muhammed MT, Aki-Yalcin E. Molecular docking: principles, advances, and its applications in drug discovery. Letters in Drug Design and Discovery. 2024 Mar 1; 21(3):480-95. doi.org/10.2174/1570180819666220922103109

31. Adelusi, T. I., Oyedele, A. Q. K., Boyenle, I. D., Ogunlana. Molecular modelling in drug discovery. Informatics in Medicine Unblocked, 2022. doi.org/10.4314/thrb. v25i3.11

32. Yang Y, Zhu Z, Wang X, Zhang X, Mu K, Shi Y, Peng C, Xu Z, Zhu W. Ligand-based approach for predicting drug targets and for virtual screening against COVID-19. Briefings in Bioinformatics. 2021 Mar; 22(2):1053-64. doi.org/10.1093/bib/bbaa422

33. Agnihotry, S., Pathak, R.K., Srivastav, A., Shukla, P.K., Gautam, B. (2020). Molecular Docking and Structure-Based Drug Design. In: Singh, D.B. (eds) Computer-Aided Drug Design. Springer, Singapore. doi.org/10.1007/978-981-15-6815-2_6

34. Vyas VK, Bhati S, Patel S, Ghate M. Structure-and ligand-based drug design methods for the modelling of antimalarial agents: a review of updates from 2012 onwards. Journal of Biomolecular Structure and Dynamics. 2022 Nov 29; 40(20):10481-506. doi.org/10.1080/07391102.2021.1932598

35. Slater O, Miller B, Kontoyianni M. Decoding protein-protein interactions: An overview. Current Topics in Medicinal Chemistry. 2020 Apr 1; 20(10):855-82. doi.org/10.2174/1568026620666200226105312

36. Shree P, Srivastava S, Chaube R, Tripathi YB. Protein-Protein docking interaction of Nephrin protein with the expression of MMP-9: an anti-inflammatory agent. Research Journal of Biotechnology. 2022 Jun; 17:6. doi.org/10.25303/1706rjbt67072

37. Agarwal S, Mehrotra RJ. An overview of molecular docking. JSM chem. 2016; 4(2): 1024-8.

38. Poudyal B. Molecular Docking Technique. Nat. J. Pharm. Sci. 2021; 1:8-14.

39. Morris, G.M., Lim-Wilby, M. (2008). Molecular Docking. In: Kukol, A. (eds) Molecular Modelling of Proteins. Methods Molecular Biology™, vol 443. Humana Press. doi.org/10.1007/978-1-59745-177-2_19

40. Yan Z, Wang J. Optimizing the affinity and specificity of ligand binding with the inclusion of solvation effect. Proteins: Structure, Function, and Bioinformatics. 2015 Sep; 83(9): 1632-42. doi.org/10.1002/prot.24848

41. Zilian D, Sotriffer CA. Sfcscore rf: a random forest-based scoring function for improved affinity prediction of protein–ligand complexes. Journal of Chemical Information and Modelling. 2013 Aug 26; 53(8):1923-33. doi.org/10.1021/ci400120b

42. Gabel J, Desaphy J, Rognan D. Beware of Machine Learning-Based Scoring Functions On the Danger of Developing Black Boxes. Journal of Chemical Information and Modelling. 2014 Oct 27; 54(10):2807-15. doi.org/10.1021/ci500406k

43. Yuriev E, Ramsland PA. Latest developments in molecular docking: 2010–2011 in review. Journal of Molecular Recognition. 2013 May; 26(5):215-39. doi.org/10.1002/jmr.2266

44. Zaki AA, Ashour A, Elhady SS, Darwish KM, Al-Karmalawy AA. Calendulaglycoside A showing potential activity against SARS-CoV-2 main protease: Molecular docking, molecular dynamics, and SAR studies. Journal of Traditional and Complementary Medicine. 2022 Jan 1; 12(1):16-34. doi.org/10.1016/j.jtcme.2021.05.001

45. Azad I. Molecular docking in the study of ligand-protein recognition: an overview. Molecular Docking-Recent Advances. 2023 Jan 25. doi.org/10.5772/intechopen.106583

46. Masters M, Mahmoud AH, Lill MA. PocketNet: ligand-guided pocket prediction for blind docking. InICLR 2023-Machine Learning for Drug Discovery workshop 2023.