BP-Dock: a flexible docking scheme for exploring protein-ligand interactions based on unbound structures

Engineering gain-of-function mutants of a WW domain by dynamics and structural analysis

Proteins gain optimal fitness such as foldability and function through evolutionary selection. However, classical studies have found that evolutionarily designed protein sequences alone cannot guarantee foldability, or at least not without considering local contacts associated with the initial folding steps. We previously showed that foldability and function can be restored by removing frustration in the folding energy landscape of a model WW domain protein, CC16, which was designed based on Statistical Coupling Analysis (SCA). Substitutions ensuring the formation of five local contacts identified as "on-path" were selected using the closest homolog native folded sequence, N21. Surprisingly, the resulting sequence, CC16-N21, bound to Group I peptides, while N21 did not. Here, we identified single-point mutations that enable N21 to bind a Group I peptide ligand through structure and dynamic-based computational design. Comparison of the docked position of the CC16-N21/ligand complex with the N21 structure showed that residues at positions 9 and 19 are important for peptide binding, whereas the dynamic profiles identified position 10 as allosterically coupled to the binding site and exhibiting different dynamics between N21 and CC16-N21. We found that swapping these positions in N21 with matched residues from CC16-N21 recovers nature-like binding affinity to N21. This study validates the use of dynamic profiles as guiding principles for affecting the binding affinity of small proteins.

Promising SARS-CoV-2 main protease inhibitor ligand-binding modes evaluated using LB-PaCS-MD/FMO

Parallel cascade selection molecular dynamics-based ligand binding-path sampling (LB-PaCS-MD) was combined with fragment molecular orbital (FMO) calculations to reveal the ligand path from an aqueous solution to the SARS-CoV-2 main protease (M^pro) active site and to customise a ligand-binding pocket suitable for delivering a potent inhibitor. Rubraxanthone exhibited mixed-inhibition antiviral activity against SARS-CoV-2 M^pro, relatively low cytotoxicity, and high cellular inhibition. However, the atomic inhibition mechanism remains ambiguous. LB-PaCS-MD/FMO is a hybrid ligand-binding evaluation method elucidating how rubraxanthone interacts with SARS-CoV-2 M^pro. In the first step, LB-PaCS-MD, which is regarded as a flexible docking, efficiently samples a set of ligand-binding pathways. After that, a reasonable docking pose of LB-PaCS-MD is evaluated by the FMO calculation to elucidate a set of protein-ligand interactions, enabling one to know the binding affinity of a specified ligand with respect to a target protein. A possible conformation was proposed for rubraxanthone binding to the SARS-CoV-2 M^pro active site, and allosteric inhibition was elucidated by combining blind docking with k-means clustering. The interaction profile, key binding residues, and considerable interaction were elucidated for rubraxanthone binding to both M^pro sites. Integrated LB-PaCS-MD/FMO provided a more reasonable complex structure for ligand binding at the SARS-CoV-2 M^pro active site, which is vital for discovering and designing antiviral drugs.

Design of novel cyanovirin-N variants by modulation of binding dynamics through distal mutations

We develop integrated co-evolution and dynamic coupling (ICDC) approach to identify, mutate, and assess distal sites to modulate function. We validate the approach first by analyzing the existing mutational fitness data of TEM-1 β-lactamase and show that allosteric positions co-evolved and dynamically coupled with the active site significantly modulate function. We further apply ICDC approach to identify positions and their mutations that can modulate binding affinity in a lectin, cyanovirin-N (CV-N), that selectively binds to dimannose, and predict binding energies of its variants through Adaptive BP-Dock. Computational and experimental analyses reveal that binding enhancing mutants identified by ICDC impact the dynamics of the binding pocket, and show that rigidification of the binding residues compensates for the entropic cost of binding. This work suggests a mechanism by which distal mutations modulate function through dynamic allostery and provides a blueprint to identify candidates for mutagenesis in order to optimize protein function.

A Coarse-Grained Methodology Identifies Intrinsic Mechanisms That Dissociate Interacting Protein Pairs

We address the problem of triggering dissociation events between proteins that have formed a complex. We have collected a set of 25 non-redundant, functionally diverse protein complexes having high-resolution three-dimensional structures in both the unbound and bound forms. We unify elastic network models with perturbation response scanning (PRS) methodology as an efficient approach for predicting residues that have the propensity to trigger dissociation of an interacting protein pair, using the three-dimensional structures of the bound and unbound proteins as input. PRS reveals that while for a group of protein pairs, residues involved in the conformational shifts are confined to regions with large motions, there are others where they originate from parts of the protein unaffected structurally by binding. Strikingly, only a few of the complexes have interface residues responsible for dissociation. We find two main modes of response: In one mode, remote control of dissociation in which disruption of the electrostatic potential distribution along protein surfaces play the major role; in the alternative mode, mechanical control of dissociation by remote residues prevail. In the former, dissociation is triggered by changes in the local environment of the protein, e.g., pH or ionic strength, while in the latter, specific perturbations arriving at the controlling residues, e.g., via binding to a third interacting partner is required for decomplexation. We resolve the observations by relying on an electromechanical coupling model which reduces to the usual elastic network result in the limit of the lack of coupling. We validate the approach by illustrating the biological significance of top residues selected by PRS on select cases where we show that the residues whose perturbation leads to the observed conformational changes correspond to either functionally important or highly conserved residues in the complex.

Atlasing of Assembly Landscapes using Distance Geometry and Graph Rigidity

This Article describes a novel geometric methodology for analyzing free energy and kinetics of assembly driven by short-range pair-potentials in an implicit solvent and provides a proof-of-concept illustration of its unique capabilities. An atlas is a labeled partition of the assembly landscape into a roadmap of maximal, contiguous, nearly-equipotential-energy conformational regions or macrostates, together with their neighborhood relationships. The new methodology decouples the roadmap generation from sampling and produces: (1) a queryable atlas of local potential energy minima, their basin structure, energy barriers, and neighboring basins; (2) paths between a specified pair of basins, each path being a sequence of conformational regions or macrostates below a desired energy threshold; and (3) approximations of relative path lengths, basin volumes (configurational entropy), and path probabilities. Results demonstrating the core algorithm's capabilities and high computational efficiency have been generated by a resource-light, curated open source software implementation EASAL (Efficient Atlasing and Search of Assembly Landscapes, ACM Trans. Math. Softw. 2018 44, 1-48. 10.1145/3204472; see software, Efficient Atlasing and Search of Assembly Landscapes, 2016. https://bitbucket.org/geoplexity/easal; video, Video Illustrating the opensource software EASAL, 2016. https://cise.ufl.edu/~sitharam/EASALvideo.mpeg; and user guide, EASAL software user guide, 2016. https://bitbucket.org/geoplexity/easal/src/master/CompleteUserGuide.pdf). Running on a laptop with Intel(R) Core(TM) i7-7700@3.60 GHz CPU with 16GB of RAM, EASAL atlases several hundred thousand conformational regions or macrostates in minutes using a single compute core. Subsequent path and basin computations each take seconds. A parallelized EASAL version running on the same laptop with 4 cores gives a 3× speedup for atlas generation. The core algorithm's correctness, time complexity, and efficiency-accuracy trade-offs are formally guaranteed using modern distance geometry, geometric constraint systems and combinatorial rigidity. The methodology further links the shape of the input assembling units to a type of intuitive and queryable bar-code of the output atlas, which in turn determine stable assembled structures and kinetics. This succinct input-output relationship facilitates reverse analysis and control toward design. A novel feature that is crucial to both the high sampling efficiency and decoupling of roadmap generation from sampling is a recently developed theory of convex Cayley (distance-based) custom parametrizations specific to assembly, as opposed to folding. Representing microstates with macrostate-specific Cayley parameters, to generate microstate samples, avoids gradient-descent search used by all prevailing methods. Further, these parametrizations convexify conformational regions or macrostates. This ratchets up sampling efficiency, significantly reducing number of repeated and discarded samples. These features of the new stand-alone methodology can also be used to complement the strengths of prevailing methodologies including Molecular Dynamics, Monte Carlo, and Fast Fourier Transform based methods.

Towards gaining sight of multiscale events: utilizing network models and normal modes in hybrid methods

With the explosion of normal mode analyses (NMAs) based on elastic network models (ENMs) in the last decade, and the proven precision of MD simulations for visualizing interactions at atomic scale, many hybrid methods have been proposed in recent years. These aim at exploiting the best of both worlds: the atomic precision of MD that often fall short of exploring time and length scales of biological interest, and the capability of ENM-NMA to predict the cooperative and often functional rearrangements of large structures and assemblies, albeit at low resolution. We present an overview of recent progress in the field with examples of successful applications highlighting the utility of such hybrid methods and pointing to emerging future directions guided by advances in experimental characterization of biomolecular systems structure and dynamics.

Ligand discrimination between active and inactive activation loop conformations of Aurora-A kinase is unmodified by phosphorylation

Activation loop phosphorylation changes the position of equilibrium between DFG-in-like and DFG-out-like conformations but not the conformational preference of inhibitors.

Structure-based drug design is commonly used to guide the development of potent and specific enzyme inhibitors. Many enzymes – such as protein kinases – adopt multiple conformations, and conformational interconversion is expected to impact on the design of small molecule inhibitors. We measured the dynamic equilibrium between DFG-in-like active and DFG-out-like inactive conformations of the activation loop of unphosphorylated Aurora-A alone, in the presence of the activator TPX2, and in the presence of kinase inhibitors. The unphosphorylated kinase had a shorter residence time of the activation loop in the active conformation and a shift in the position of equilibrium towards the inactive conformation compared with phosphorylated kinase for all conditions measured. Ligand binding was associated with a change in the position of conformational equilibrium which was specific to each ligand and independent of the kinase phosphorylation state. As a consequence of this, the ability of a ligand to discriminate between active and inactive activation loop conformations was also independent of phosphorylation. Importantly, we discovered that the presence of multiple enzyme conformations can lead to a plateau in the overall ligand K_d, despite increasing affinity for the chosen target conformation, and modelled the conformational discrimination necessary for a conformation-promoting ligand.

Enriching Traditional Protein-protein Interaction Networks with Alternative Conformations of Proteins

Traditional Protein-Protein Interaction (PPI) networks, which use a node and edge representation, lack some valuable information about the mechanistic details of biological processes. Mapping protein structures to these PPI networks not only provides structural details of each interaction but also helps us to find the mutual exclusive interactions. Yet it is not a comprehensive representation as it neglects the conformational changes of proteins which may lead to different interactions, functions, and downstream signalling. In this study, we proposed a new representation for structural PPI networks inspecting the alternative conformations of proteins. We performed a large-scale study by creating breast cancer metastasis network and equipped it with different conformers of proteins. Our results showed that although 88% of proteins in our network has at least two structures in Protein Data Bank (PDB), only 22% of them have alternative conformations and the remaining proteins have different regions saved in PDB. However, using even this small set of alternative conformations we observed a considerable increase in our protein docking predictions. Our protein-protein interaction predictions increased from 54% to 76% using the alternative conformations. We also showed the benefits of investigating structural data and alternative conformations of proteins through three case studies.

The Development of Target-Specific Pose Filter Ensembles To Boost Ligand Enrichment for Structure-Based Virtual Screening

Structure-based virtual screening (SBVS) has become an indispensable technique for hit identification at the early stage of drug discovery. However, the accuracy of current scoring functions is not high enough to confer success to every target and thus remains to be improved. Previously, we had developed binary pose filters (PFs) using knowledge derived from the protein-ligand interface of a single X-ray structure of a specific target. This novel approach had been validated as an effective way to improve ligand enrichment. Continuing from it, in the present work we attempted to incorporate knowledge collected from diverse protein-ligand interfaces of multiple crystal structures of the same target to build PF ensembles (PFEs). Toward this end, we first constructed a comprehensive data set to meet the requirements of ensemble modeling and validation. This set contains 10 diverse targets, 118 well-prepared X-ray structures of protein-ligand complexes, and large benchmarking actives/decoys sets. Notably, we designed a unique workflow of two-layer classifiers based on the concept of ensemble learning and applied it to the construction of PFEs for all of the targets. Through extensive benchmarking studies, we demonstrated that (1) coupling PFE with Chemgauss4 significantly improves the early enrichment of Chemgauss4 itself and (2) PFEs show greater consistency in boosting early enrichment and larger overall enrichment than our prior PFs. In addition, we analyzed the pairwise topological similarities among cognate ligands used to construct PFEs and found that it is the higher chemical diversity of the cognate ligands that leads to the improved performance of PFEs. Taken together, the results so far prove that the incorporation of knowledge from diverse protein-ligand interfaces by ensemble modeling is able to enhance the screening competence of SBVS scoring functions.

A better prediction of conformational changes of proteins using minimally connected network models

Elastic network models have recently been used for studying low-frequency collective motions of proteins. These models simplify the complexity that arises from normal mode analysis by considering a simplified potential involving a few parameters. Two common parameters in most of the elastic network models are cutoff radius and force constant. Although the latter has been studied extensively and even elaborate new models were introduced, for the former usually an ad-hoc cutoff radius is considered. Moreover, the quality of the network models is usually assessed by evaluating their prediction against experimental B-factors. In this work, we consider various common elastic network models with different cutoff radii and assess them by their ability to predict conformational changes of proteins in complexes from unbound to bound state. This prediction is performed by perturbing the unbound structure using a number of low-frequency normal modes of its network model to optimally fit the bound structure. We evaluated a dataset of 30 proteins with distinct unbound and bound structures using this criterion. The results showed that, opposed to the common calibration process based on B-factors, a meaningful relationship exists between the quality of the prediction and model parameters. It was shown that the cutoff radius has a major role in this prediction and minimally connected network models, which use the shortest cutoff radius for which the network is stable, give the best results. It was also shown that by considering the first ten normal modes, the conformational changes can be predicted by about 25 percent. Hence, the evaluation process was extended to the case of considering the contribution of all normal modes in the prediction. The results indicated that minimally connected network models are superior, despite their simplicity, when any number of modes are considered in the prediction.

Sequence-Based Prediction of Protein-Carbohydrate Binding Sites Using Support Vector Machines

Carbohydrate-binding proteins play significant roles in many diseases including cancer. Here, we established a machine-learning-based method (called sequence-based prediction of residue-level interaction sites of carbohydrates, SPRINT-CBH) to predict carbohydrate-binding sites in proteins using support vector machines (SVMs). We found that integrating evolution-derived sequence profiles with additional information on sequence and predicted solvent accessible surface area leads to a reasonably accurate, robust, and predictive method, with area under receiver operating characteristic curve (AUC) of 0.78 and 0.77 and Matthew's correlation coefficient of 0.34 and 0.29, respectively for 10-fold cross validation and independent test without balancing binding and nonbinding residues. The quality of the method is further demonstrated by having statistically significantly more binding residues predicted for carbohydrate-binding proteins than presumptive nonbinding proteins in the human proteome, and by the bias of rare alleles toward predicted carbohydrate-binding sites for nonsynonymous mutations from the 1000 genome project. SPRINT-CBH is available as an online server at http://sparks-lab.org/server/SPRINT-CBH .

Exploring Multiple Binding Modes Using Confined Replica Exchange Molecular Dynamics

Molecular docking is extensively applied to determine the position of a ligand on its receptor despite the rather poor correspondence between docking scores and experimental binding affinities found in several studies, especially for systems structurally unrelated with those used in the scoring functions' training sets. Here, we present a method for the prediction of binding modes and binding free energies, which uses replica exchange molecular dynamics in combination with a receptor-shaped piecewise potential, confining the ligand in the proximity of the receptor surface and limiting the accessible conformational space of interest. We assess our methodology with a set of protein receptor-ligand test cases. In every case studied, the method is able to locate the ligand on the experimentally known receptor binding site, and it gives as output the binding free energy. The added value of our approach with respect to other available methods is that it quickly performs a conformational space search, providing a set of bound (or unbound) configurations, which can be used to determine phenomenological structural and energetic properties of an experimental binding state as a result of contributions provided by diversified multiple binding poses.

Human Aquaporin-4 and Molecular Modeling: Historical Perspective and View to the Future

Among the different aquaporins (AQPs), human aquaporin-4 (hAQP4) has attracted the greatest interest in recent years as a new promising therapeutic target. Such a membrane protein is, in fact, involved in a multiple sclerosis-like immunopathology called Neuromyelitis Optica (NMO) and in several disorders resulting from imbalanced water homeostasis such as deafness and cerebral edema. The gap of knowledge in its functioning and dynamics at the atomistic level of detail has hindered the development of rational strategies for designing hAQP4 modulators. The application, lately, of molecular modeling has proved able to fill this gap providing a breeding ground to rationally address compounds targeting hAQP4. In this review, we give an overview of the important advances obtained in this field through the application of Molecular Dynamics (MD) and other complementary modeling techniques. The case studies presented herein are discussed with the aim of providing important clues for computational chemists and biophysicists interested in this field and looking for new challenges.

Ligand Docking to Intermediate and Close-To-Bound Conformers Generated by an Elastic Network Model Based Algorithm for Highly Flexible Proteins

Incorporating receptor flexibility in small ligand-protein docking still poses a challenge for proteins undergoing large conformational changes. In the absence of bound structures, sampling conformers that are accessible by apo state may facilitate docking and drug design studies. For this aim, we developed an unbiased conformational search algorithm, by integrating global modes from elastic network model, clustering and energy minimization with implicit solvation. Our dataset consists of five diverse proteins with apo to complex RMSDs 4.7-15 Å. Applying this iterative algorithm on apo structures, conformers close to the bound-state (RMSD 1.4-3.8 Å), as well as the intermediate states were generated. Dockings to a sequence of conformers consisting of a closed structure and its "parents" up to the apo were performed to compare binding poses on different states of the receptor. For two periplasmic binding proteins and biotin carboxylase that exhibit hinge-type closure of two dynamics domains, the best pose was obtained for the conformer closest to the bound structure (ligand RMSDs 1.5-2 Å). In contrast, the best pose for adenylate kinase corresponded to an intermediate state with partially closed LID domain and open NMP domain, in line with recent studies (ligand RMSD 2.9 Å). The docking of a helical peptide to calmodulin was the most challenging case due to the complexity of its 15 Å transition, for which a two-stage procedure was necessary. The technique was first applied on the extended calmodulin to generate intermediate conformers; then peptide docking and a second generation stage on the complex were performed, which in turn yielded a final peptide RMSD of 2.9 Å. Our algorithm is effective in producing conformational states based on the apo state. This study underlines the importance of such intermediate states for ligand docking to proteins undergoing large transitions.

Adaptive BP-Dock: An Induced Fit Docking Approach for Full Receptor Flexibility

We present an induced fit docking approach called Adaptive BP-Dock that integrates perturbation response scanning (PRS) with the flexible docking protocol of RosettaLigand in an adaptive manner. We first perturb the binding pocket residues of a receptor and obtain a new conformation based on the residue response fluctuation profile using PRS. Next, we dock a ligand to this new conformation by RosettaLigand, where we repeat these steps for several iterations. We test this approach on several protein test sets including difficult unbound docking cases such as HIV-1 reverse transcriptase and HIV-1 protease. Adaptive BP-Dock results show better correlation with experimental binding affinities compared to other docking protocols. Overall, the results imply that Adaptive BP-Dock can easily capture binding induced conformational changes by simultaneous sampling of protein and ligand conformations. This can provide faster and efficient docking of novel targets for rational drug design.

Predicting long term cooperativity and specific modulators of receptor interactions in human transferrin from dynamics within a single microstate

PRS identifies regions contacting rapidly evolving residues that mechanically manipulate dissociation from the pathogen in the human transferrin–bacterial receptor complex.

A Rigid Hinge Region Is Necessary for High-Affinity Binding of Dimannose to Cyanovirin and Associated Constructs

Mutations in the hinge region of cyanovirin-N (CVN) dictate its preferential oligomerization state. Constructs with the Pro51Gly mutation preferentially exist as monomers, whereas wild-type cyanovirin can form domain-swapped dimers under certain conditions. Because the hinge region is an integral part of the high-affinity binding site of CVN, we investigated whether this mutation affects the shape, flexibility, and binding affinity of domain B for dimannose. Our studies indicate that the capability of monomeric wild-type CVN to resist mechanical perturbations is enhanced when compared to that of constructs in which the hinge region is more flexible. Our computational results also show that enhanced flexibility leads to blocking of the binding site by allowing different rotational isomeric states of Asn53. Moreover, at higher temperatures, this observed flexibility leads to an interaction between Asn53 and Asn42, further hindering access to the binding site. On the basis of these results, we predicted that binding affinity for dimannose would be more favorable for cyanovirin constructs containing a wild-type hinge region, whereas affinity would be impaired in the case of mutants containing Pro51Gly. Experimental characterization by isothermal titration calorimetry of a set of cyanovirin mutants confirms this hypothesis. Those possessing the Pro51Gly mutation are consistently inferior binders.

Sampling of conformational ensemble for virtual screening using molecular dynamics simulations and normal mode analysis

Aim: Molecular dynamics simulations and normal mode analysis are well-established approaches to generate receptor conformational ensembles (RCEs) for ligand docking and virtual screening. Here, we report new fast molecular dynamics-based and normal mode analysis-based protocols combined with conformational pocket classifications to efficiently generate RCEs. Materials & Methods: We assessed our protocols on two well-characterized protein targets showing local active site flexibility, dihydrofolate reductase and large collective movements, CDK2. The performance of the RCEs was validated by distinguishing known ligands of dihydrofolate reductase and CDK2 among a dataset of diverse chemical decoys. Results & discussion: Our results show that different simulation protocols can be efficient for generation of RCEs depending on different kind of protein flexibility.[Formula: see text]

The Role of Conformational Dynamics and Allostery in the Disease Development of Human Ferritin

Determining the three-dimensional structure of myoglobin, the first solved structure of a protein, fundamentally changed the way protein function was understood. Even more revolutionary was the information that came afterward: protein dynamics play a critical role in biological functions. Therefore, understanding conformational dynamics is crucial to obtaining a more complete picture of protein evolution. We recently analyzed the evolution of different protein families including green fluorescent proteins (GFPs), β-lactamase inhibitors, and nuclear receptors, and we observed that the alteration of conformational dynamics through allosteric regulation leads to functional changes. Moreover, proteome-wide conformational dynamics analysis of more than 100 human proteins showed that mutations occurring at rigid residue positions are more susceptible to disease than flexible residue positions. These studies suggest that disease-associated mutations may impair dynamic allosteric regulations, leading to loss of function. Thus, in this study, we analyzed the conformational dynamics of the wild-type light chain subunit of human ferritin protein along with the neutral and disease forms. We first performed replica exchange molecular dynamics simulations of wild-type and mutants to obtain equilibrated dynamics and then used perturbation response scanning (PRS), where we introduced a random Brownian kick to a position and computed the fluctuation response of the chain using linear response theory. Using this approach, we computed the dynamic flexibility index (DFI) for each position in the chain for the wild-type and the mutants. DFI quantifies the resilience of a position to a perturbation and provides a flexibility/rigidity measurement for a given position in the chain. The DFI analysis reveals that neutral variants and the wild-type exhibit similar flexibility profiles in which experimentally determined functionally critical sites act as hinges in controlling the overall motion. However, disease mutations alter the conformational dynamic profile, making hinges more loose (i.e., softening the hinges), thus impairing the allosterically regulated dynamics.

Role of computer-aided drug design in modern drug discovery

Drug discovery utilizes chemical biology and computational drug design approaches for the efficient identification and optimization of lead compounds. Chemical biology is mostly involved in the elucidation of the biological function of a target and the mechanism of action of a chemical modulator. On the other hand, computer-aided drug design makes use of the structural knowledge of either the target (structure-based) or known ligands with bioactivity (ligand-based) to facilitate the determination of promising candidate drugs. Various virtual screening techniques are now being used by both pharmaceutical companies and academic research groups to reduce the cost and time required for the discovery of a potent drug. Despite the rapid advances in these methods, continuous improvements are critical for future drug discovery tools. Advantages presented by structure-based and ligand-based drug design suggest that their complementary use, as well as their integration with experimental routines, has a powerful impact on rational drug design. In this article, we give an overview of the current computational drug design and their application in integrated rational drug development to aid in the progress of drug discovery research.

Challenges, applications, and recent advances of protein-ligand docking in structure-based drug design

The docking methods used in structure-based virtual database screening offer the ability to quickly and cheaply estimate the affinity and binding mode of a ligand for the protein receptor of interest, such as a drug target. These methods can be used to enrich a database of compounds, so that more compounds that are subsequently experimentally tested are found to be pharmaceutically interesting. In addition, like all virtual screening methods used for drug design, structure-based virtual screening can focus on curated libraries of synthesizable compounds, helping to reduce the expense of subsequent experimental verification. In this review, we introduce the protein-ligand docking methods used for structure-based drug design and other biological applications. We discuss the fundamental challenges facing these methods and some of the current methodological topics of interest. We also discuss the main approaches for applying protein-ligand docking methods. We end with a discussion of the challenging aspects of evaluating or benchmarking the accuracy of docking methods for their improvement, and discuss future directions.