Targeted scoring functions for virtual screening

SAnDReS 2.0: Development of machine-learning models to explore the scoring function space

Classical scoring functions may exhibit low accuracy in determining ligand binding affinity for proteins. The availability of both protein-ligand structures and affinity data make it possible to develop machine-learning models focused on specific protein systems with superior predictive performance. Here, we report a new methodology named SAnDReS that combines AutoDock Vina 1.2 with 54 regression methods available in Scikit-Learn to calculate binding affinity based on protein-ligand structures. This approach allows exploration of the scoring function space. SAnDReS generates machine-learning models based on crystal, docked, and AlphaFold-generated structures. As a proof of concept, we examine the performance of SAnDReS-generated models in three case studies. For all three cases, our models outperformed classical scoring functions. Also, SAnDReS-generated models showed predictive performance close to or better than other machine-learning models such as KDEEP, CSM-lig, and ΔVinaRF20. SAnDReS 2.0 is available to download at https://github.com/azevedolab/sandres.

Molecular Dynamics as a Tool for Virtual Ligand Screening

Rational drug design is essential for new drugs to emerge, especially when the structure of a target protein or nucleic acid is known. To that purpose, high-throughput virtual ligand screening campaigns aim at discovering computationally new binding molecules or fragments to modulate particular biomolecular interactions or biological activities, related to a disease process. The structure-based virtual ligand screening process primarily relies on docking methods which allow predicting the binding of a molecule to a biological target structure with a correct conformation and the best possible affinity. The docking method itself is not sufficient as it suffers from several and crucial limitations (lack of full protein flexibility information, no solvation and ion effects, poor scoring functions, and unreliable molecular affinity estimation).At the interface of computer techniques and drug discovery, molecular dynamics (MD) allows introducing protein flexibility before or after a docking protocol, refining the structure of protein-drug complexes in the presence of water, ions, and even in membrane-like environments, describing more precisely the temporal evolution of the biological complex and ranking these complexes with more accurate binding energy calculations. In this chapter, we describe the up-to-date MD, which plays the role of supporting tools in the virtual ligand screening (VS) process.Without a doubt, using docking in combination with MD is an attractive approach in structure-based drug discovery protocols nowadays. It has proved its efficiency through many examples in the literature and is a powerful method to significantly reduce the amount of required wet experimentations (Tarcsay et al, J Chem Inf Model 53:2990-2999, 2013; Barakat et al, PLoS One 7:e51329, 2012; De Vivo et al, J Med Chem 59:4035-4061, 2016; Durrant, McCammon, BMC Biol 9:71-79, 2011; Galeazzi, Curr Comput Aided Drug Des 5:225-240, 2009; Hospital et al, Adv Appl Bioinforma Chem 8:37-47, 2015; Jiang et al, Molecules 20:12769-12786, 2015; Kundu et al, J Mol Graph Model 61:160-174, 2015; Mirza et al, J Mol Graph Model 66:99-107, 2016; Moroy et al, Future Med Chem 7:2317-2331, 2015; Naresh et al, J Mol Graph Model 61:272-280, 2015; Nichols et al, J Chem Inf Model 51:1439-1446, 2011; Nichols et al, Methods Mol Biol 819:93-103, 2012; Okimoto et al, PLoS Comput Biol 5:e1000528, 2009; Rodriguez-Bussey et al, Biopolymers 105:35-42, 2016; Sliwoski et al, Pharmacol Rev 66:334-395, 2014).

Target-ligand binding affinity from single point enthalpy calculation and elemental composition

Reliable target-ligand binding thermodynamics data are essential for successful drug design and molecular engineering projects. Besides experimental methods, a number of theoretical approaches have been introduced for the generation of binding thermodynamics data. However, available approaches often neglect electronic effects or explicit water molecules influencing target-ligand interactions. To handle electronic effects within a reasonable time frame, we introduce a fast calculator QMH-L using a single target-ligand complex structure pre-optimized at the molecular mechanics level. QMH-L is composed of the semi-empirical quantum mechanics calculation of binding enthalpy with predicted explicit water molecules at the complex interface, and a simple descriptor based on the elemental composition of the ligand. QMH-L estimates the target-ligand binding free energy with a root mean square error (RMSE) of 0.94 kcal mol-1. The calculations also provide binding enthalpy values and they were compared with experimental binding thermodynamics data collected from the most reliable isothermal titration calorimetry studies of systems including various protein targets and challenging, large peptide ligands with a molecular weight of up to 2-3 thousand. The single point enthalpy calculations of QMH-L require modest computational resources and are based on short runs with open source and/or free software like Gromacs, Mopac, MobyWat, and Fragmenter. QMH-L can be applied for fast, automated scoring of drug candidates during a virtual screen, enthalpic engineering of new ligands or thermodynamic explanation of complex interactions.

Structural interaction fingerprints and machine learning for predicting and explaining binding of small molecule ligands to RNA

Ribonucleic acids (RNAs) play crucial roles in living organisms and some of them, such as bacterial ribosomes and precursor messenger RNA, are targets of small molecule drugs, whereas others, e.g. bacterial riboswitches or viral RNA motifs are considered as potential therapeutic targets. Thus, the continuous discovery of new functional RNA increases the demand for developing compounds targeting them and for methods for analyzing RNA-small molecule interactions. We recently developed fingeRNAt-a software for detecting non-covalent bonds formed within complexes of nucleic acids with different types of ligands. The program detects several non-covalent interactions and encodes them as structural interaction fingerprint (SIFt). Here, we present the application of SIFts accompanied by machine learning methods for binding prediction of small molecules to RNA. We show that SIFt-based models outperform the classic, general-purpose scoring functions in virtual screening. We also employed Explainable Artificial Intelligence (XAI)-the SHapley Additive exPlanations, Local Interpretable Model-agnostic Explanations and other methods to help understand the decision-making process behind the predictive models. We conducted a case study in which we applied XAI on a predictive model of ligand binding to human immunodeficiency virus type 1 trans-activation response element RNA to distinguish between residues and interaction types important for binding. We also used XAI to indicate whether an interaction has a positive or negative effect on binding prediction and to quantify its impact. Our results obtained using all XAI methods were consistent with the literature data, demonstrating the utility and importance of XAI in medicinal chemistry and bioinformatics.

Application of Machine Learning Techniques to Predict Binding Affinity for Drug Targets: A Study of Cyclin-Dependent Kinase 2

BACKGROUND

The elucidation of the structure of cyclin-dependent kinase 2 (CDK2) made it possible to develop targeted scoring functions for virtual screening aimed to identify new inhibitors for this enzyme. CDK2 is a protein target for the development of drugs intended to modulate cellcycle progression and control. Such drugs have potential anticancer activities.

OBJECTIVE

Our goal here is to review recent applications of machine learning methods to predict ligand- binding affinity for protein targets. To assess the predictive performance of classical scoring functions and targeted scoring functions, we focused our analysis on CDK2 structures.

METHODS

We have experimental structural data for hundreds of binary complexes of CDK2 with different ligands, many of them with inhibition constant information. We investigate here computational methods to calculate the binding affinity of CDK2 through classical scoring functions and machine- learning models.

RESULTS

Analysis of the predictive performance of classical scoring functions available in docking programs such as Molegro Virtual Docker, AutoDock4, and Autodock Vina indicated that these methods failed to predict binding affinity with significant correlation with experimental data. Targeted scoring functions developed through supervised machine learning techniques showed a significant correlation with experimental data.

CONCLUSION

Here, we described the application of supervised machine learning techniques to generate a scoring function to predict binding affinity. Machine learning models showed superior predictive performance when compared with classical scoring functions. Analysis of the computational models obtained through machine learning could capture essential structural features responsible for binding affinity against CDK2.

New machine learning and physics-based scoring functions for drug discovery

Scoring functions are essential for modern in silico drug discovery. However, the accurate prediction of binding affinity by scoring functions remains a challenging task. The performance of scoring functions is very heterogeneous across different target classes. Scoring functions based on precise physics-based descriptors better representing protein–ligand recognition process are strongly needed. We developed a set of new empirical scoring functions, named DockTScore, by explicitly accounting for physics-based terms combined with machine learning. Target-specific scoring functions were developed for two important drug targets, proteases and protein–protein interactions, representing an original class of molecules for drug discovery. Multiple linear regression (MLR), support vector machine and random forest algorithms were employed to derive general and target-specific scoring functions involving optimized MMFF94S force-field terms, solvation and lipophilic interactions terms, and an improved term accounting for ligand torsional entropy contribution to ligand binding. DockTScore scoring functions demonstrated to be competitive with the current best-evaluated scoring functions in terms of binding energy prediction and ranking on four DUD-E datasets and will be useful for in silico drug design for diverse proteins as well as for specific targets such as proteases and protein–protein interactions. Currently, the MLR DockTScore is available at www.dockthor.lncc.br.

Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions

How to accurately estimate protein-ligand binding affinity remains a key challenge in computer-aided drug design (CADD). In many cases, it has been shown that the binding affinities predicted by classical scoring functions (SFs) cannot correlate well with experimentally measured biological activities. In the past few years, machine learning (ML)-based SFs have gradually emerged as potential alternatives and outperformed classical SFs in a series of studies. In this study, to better recognize the potential of classical SFs, we have conducted a comparative assessment of 25 commonly used SFs. Accordingly, the scoring power was systematically estimated by using the state-of-the-art ML methods that replaced the original multiple linear regression method to refit individual energy terms. The results show that the newly-developed ML-based SFs consistently performed better than classical ones. In particular, gradient boosting decision tree (GBDT) and random forest (RF) achieved the best predictions in most cases. The newly-developed ML-based SFs were also tested on another benchmark modified from PDBbind v2007, and the impacts of structural and sequence similarities were evaluated. The results indicated that the superiority of the ML-based SFs could be fully guaranteed when sufficient similar targets were contained in the training set. Moreover, the effect of the combinations of features from multiple SFs was explored, and the results indicated that combining NNscore2.0 with one to four other classical SFs could yield the best scoring power. However, it was not applicable to derive a generic target-specific SF or SF combination.

Molecular modeling provides a structural basis for PERK inhibitor selectivity towards RIPK1

Molecular modelling explains the lack of selectivity for inhibitors GSK2606414 and GSK2656157, as compared to inhibitor AMG44.

Predictive Power of In Silico Approach to Evaluate Chemicals against M. tuberculosis: A Systematic Review

Mycobacterium tuberculosis (Mtb) is an endemic bacterium worldwide that causes tuberculosis (TB) and involves long-term treatment that is not always effective. In this context, several studies are trying to develop and evaluate new substances active against Mtb. In silico techniques are often used to predict the effects on some known target. We used a systematic approach to find and evaluate manuscripts that applied an in silico technique to find antimycobacterial molecules and tried to prove its predictive potential by testing them in vitro or in vivo. After searching three different databases and applying exclusion criteria, we were able to retrieve 46 documents. We found that they all follow a similar screening procedure, but few studies exploited equal targets, exploring the interaction of multiple ligands to 29 distinct enzymes. The following in vitro/vivo analysis showed that, although the virtual assays were able to decrease the number of molecules tested, saving time and money, virtual screening procedures still need to develop the correlation to more favorable in vitro outcomes. We find that the in silico approach has a good predictive power for in vitro results, but call for more studies to evaluate its clinical predictive possibilities.

A Structure-Based Drug Discovery Paradigm

Structure-based drug design is becoming an essential tool for faster and more cost-efficient lead discovery relative to the traditional method. Genomic, proteomic, and structural studies have provided hundreds of new targets and opportunities for future drug discovery. This situation poses a major problem: the necessity to handle the “big data” generated by combinatorial chemistry. Artificial intelligence (AI) and deep learning play a pivotal role in the analysis and systemization of larger data sets by statistical machine learning methods. Advanced AI-based sophisticated machine learning tools have a significant impact on the drug discovery process including medicinal chemistry. In this review, we focus on the currently available methods and algorithms for structure-based drug design including virtual screening and de novo drug design, with a special emphasis on AI- and deep-learning-based methods used for drug discovery.

Empirical Scoring Functions for Structure-Based Virtual Screening: Applications, Critical Aspects, and Challenges

Structure-based virtual screening (VS) is a widely used approach that employs the knowledge of the three-dimensional structure of the target of interest in the design of new lead compounds from large-scale molecular docking experiments. Through the prediction of the binding mode and affinity of a small molecule within the binding site of the target of interest, it is possible to understand important properties related to the binding process. Empirical scoring functions are widely used for pose and affinity prediction. Although pose prediction is performed with satisfactory accuracy, the correct prediction of binding affinity is still a challenging task and crucial for the success of structure-based VS experiments. There are several efforts in distinct fronts to develop even more sophisticated and accurate models for filtering and ranking large libraries of compounds. This paper will cover some recent successful applications and methodological advances, including strategies to explore the ligand entropy and solvent effects, training with sophisticated machine-learning techniques, and the use of quantum mechanics. Particular emphasis will be given to the discussion of critical aspects and further directions for the development of more accurate empirical scoring functions.

Activity-based classification circumvents affinity prediction problems for pyrrolidine carboxamide inhibitors of InhA

Developing reliable structure-based activity prediction models for a particular ligand series can be challenging if the target is flexible and the affinity range of the training compounds is narrow. For a data set of 44 pyrrolidine carboxamide inhibitors of the mycobacterial enoyl-ACP-reductase InhA this proved to be case, as scoring methods of various origin and complexity did not succeed in providing practically useful correlations with experimental inhibition data. In contrast, logistic regression models for activity-based classification trained with combinations of scoring functions led to good separation of the more active inhibitors from the weakest compounds. The approach is suggested as an alternative in cases where classical scoring and ranking procedures fail.

Molecular Dynamics as a Tool for Virtual Ligand Screening

Rational drug design is essential for new drugs to emerge, especially when the structure of a target protein or catalytic enzyme is known experimentally. To that purpose, high-throughput virtual ligand screening campaigns aim at discovering computationally new binding molecules or fragments to inhibit a particular protein interaction or biological activity. The virtual ligand screening process often relies on docking methods which allow predicting the binding of a molecule into a biological target structure with a correct conformation and the best possible affinity. The docking method itself is not sufficient as it suffers from several and crucial limitations (lack of protein flexibility information, no solvation effects, poor scoring functions, and unreliable molecular affinity estimation).At the interface of computer techniques and drug discovery, molecular dynamics (MD) allows introducing protein flexibility before or after a docking protocol, refining the structure of protein-drug complexes in the presence of water, ions and even in membrane-like environments, and ranking complexes with more accurate binding energy calculations. In this chapter we describe the up-to-date MD protocols that are mandatory supporting tools in the virtual ligand screening (VS) process. Using docking in combination with MD is one of the best computer-aided drug design protocols nowadays. It has proved its efficiency through many examples, described below.

Strategies of Virtual Screening in Medicinal Chemistry

Virtual screening represents an effective computational strategy to rise-up the chances of finding new bioactive compounds by accelerating the time needed to move from an initial intuition to market. Classically, the most pursued approaches rely on ligand- and structure-based studies, the former employed when structural data information about the target is missing while the latter employed when X-ray/NMR solved or homology models are instead available for the target. The authors will focus on the most advanced techniques applied in this area. In particular, they will survey the key concepts of virtual screening by discussing how to properly select chemical libraries, how to make database curation, how to applying and- and structure-based techniques, how to wisely use post-processing methods. Emphasis will be also given to the most meaningful databases used in VS protocols. For the ease of discussion several examples will be presented.

Enhance the performance of current scoring functions with the aid of 3D protein-ligand interaction fingerprints

BACKGROUND

In structure-based drug design, binding affinity prediction remains as a challenging goal for current scoring functions. Development of target-biased scoring functions provides a new possibility for tackling this problem, but this approach is also associated with certain technical difficulties. We previously reported the Knowledge-Guided Scoring (KGS) method as an alternative approach (BMC Bioinformatics, 2010, 11, 193-208). The key idea is to compute the binding affinity of a given protein-ligand complex based on the known binding data of an appropriate reference complex, so the error in binding affinity prediction can be reduced effectively.

RESULTS

In this study, we have developed an upgraded version, i.e. KGS2, by employing 3D protein-ligand interaction fingerprints in reference selection. KGS2 was evaluated in combination with four scoring functions (X-Score, ChemPLP, ASP, and GoldScore) on five drug targets (HIV-1 protease, carbonic anhydrase 2, beta-secretase 1, beta-trypsin, and checkpoint kinase 1). In the in situ scoring test, considerable improvements were observed in most cases after application of KGS2. Besides, the performance of KGS2 was always better than KGS in all cases. In the more challenging molecular docking test, application of KGS2 also led to improved structure-activity relationship in some cases.

CONCLUSIONS

KGS2 can be applied as a convenient "add-on" to current scoring functions without the need to re-engineer them, and its application is not limited to certain target proteins as customized scoring functions. As an interpolation method, its accuracy in principle can be improved further with the increasing knowledge of protein-ligand complex structures and binding affinity data. We expect that KGS2 will become a practical tool for enhancing the performance of current scoring functions in binding affinity prediction. The KGS2 software is available upon contacting the authors.

Scoring Functions of Protein-Ligand Interactions

Scoring function of protein-ligand interactions is used to recognize the “native” binding pose of a ligand on the protein and to predict the binding affinity, so that the active small molecules can be discriminated from the non-active ones. Scoring function is widely used in computationally molecular docking and structure-based drug discovery. The development and improvement of scoring functions have broad implications in pharmaceutical industry and academic research. During the past three decades, much progress have been made in methodology and accuracy for scoring functions, and many successful cases have be witnessed in virtual database screening. In this chapter, the authors introduced the basic types of scoring functions and their derivations, the commonly-used evaluation methods and benchmarks, as well as the underlying challenges and current solutions. Finally, the authors discussed the promising directions to improve and develop scoring functions for future molecular docking-based drug discovery.

Empirical Scoring Functions for Affinity Prediction of Protein-ligand Complexes

The ability to rapidly assess the quality of a protein-ligand complex in terms of its affinity is of fundamental importance for various methods of computer-aided drug design. While simple filtering or matching critieria may be sufficient in fast docking methods or at early stages of virtual screening, estimates of the actual free energy of binding are needed whenever refined docking solutions, ligand rankings or support for the optimization of hit compounds are required. If rigorous free energy calculations based on molecular simulations are impractical, such affinity estimates are provided by scoring functions. The class of empirical scoring functions aims to provide them via a regression-based approach. Using experimental structures and affinity data of protein-ligand complexes and descriptors suitable to capture the essential features of the interaction, these functions are trained with classical linear regression techniques or machine-learning methods. The latter have led to considerable improvements in terms of prediction accuracy for large generic data sets. Nevertheless, many limitations are not yet resolved and pose significant challenges for future developments.

Reliability analysis and optimization of the consensus docking approach for the development of virtual screening studies

Ligand-protein docking is one of the most common techniques used in virtual screening campaigns. Despite the large number of docking software available, there is still the need of improving the efficacy of docking-based virtual screenings. To date, only very few studies evaluated the possibility of combining the results of different docking methods to achieve higher success rates in virtual screening studies (consensus docking). In order to better understand the range of applicability of this approach, we carried out an extensive enriched database analysis using the DUD dataset. The consensus docking protocol was then refined by applying modifications concerning the calculation of pose consensus and the combination of docking methods included in the procedure. The results obtained suggest that this approach performs as well as the best available methods found in literature, confirming the idea that this procedure can be profitably used for the identification of new hit compounds.

Applying DEKOIS 2.0 in structure-based virtual screening to probe the impact of preparation procedures and score normalization

Background

Structure-based virtual screening techniques can help to identify new lead structures and complement other screening approaches in drug discovery. Prior to docking, the data (protein crystal structures and ligands) should be prepared with great attention to molecular and chemical details.

Results

Using a subset of 18 diverse targets from the recently introduced DEKOIS 2.0 benchmark set library, we found differences in the virtual screening performance of two popular docking tools (GOLD and Glide) when employing two different commercial packages (e.g. MOE and Maestro) for preparing input data. We systematically investigated the possible factors that can be responsible for the found differences in selected sets. For the Angiotensin-I-converting enzyme dataset, preparation of the bioactive molecules clearly exerted the highest influence on VS performance compared to preparation of the decoys or the target structure. The major contributing factors were different protonation states, molecular flexibility, and differences in the input conformation (particularly for cyclic moieties) of bioactives. In addition, score normalization strategies eliminated the biased docking scores shown by GOLD (ChemPLP) for the larger bioactives and produced a better performance. Generalizing these normalization strategies on the 18 DEKOIS 2.0 sets, improved the performances for the majority of GOLD (ChemPLP) docking, while it showed detrimental performances for the majority of Glide (SP) docking.

Conclusions

In conclusion, we exemplify herein possible issues particularly during the preparation stage of molecular data and demonstrate to which extent these issues can cause perturbations in the virtual screening performance. We provide insights into what problems can occur and should be avoided, when generating benchmarks to characterize the virtual screening performance. Particularly, careful selection of an appropriate molecular preparation setup for the bioactive set and the use of score normalization for docking with GOLD (ChemPLP) appear to have a great importance for the screening performance. For virtual screening campaigns, we recommend to invest time and effort into including alternative preparation workflows into the generation of the master library, even at the cost of including multiple representations of each molecule.

Graphical Abstract

Evaluation of 11 scoring functions performance on matrix metalloproteinases

Matrix metalloproteinases (MMPs) have distinctive roles in various physiological and pathological processes such as inflammatory diseases and cancer. This study explored the performance of eleven scoring functions (D-Score, G-Score, ChemScore, F-Score, PMF-Score, PoseScore, RankScore, DSX, and X-Score and scoring functions of AutoDock4.1 and AutoDockVina). Their performance was judged by calculation of their correlations to experimental binding affinities of 3D ligand-enzyme complexes of MMP family. Furthermore, they were evaluated for their ability in reranking virtual screening study results performed on a member of MMP family (MMP-12). Enrichment factor at different levels and receiver operating characteristics (ROC) curves were used to assess their performance. Finally, we have developed a PCA model from the best functions. Of the scoring functions evaluated, F-Score, DSX, and ChemScore were the best overall performers in prediction of MMPs-inhibitors binding affinities while ChemScore, Autodock, and DSX had the best discriminative power in virtual screening against the MMP-12 target. Consensus scorings did not show statistically significant superiority over the other scorings methods in correlation study while PCA model which consists of ChemScore, Autodock, and DSX improved overall enrichment. Outcome of this study could be useful for the setting up of a suitable scoring protocol, resulting in enrichment of MMPs inhibitors.

Comparative analysis of various electrostatic potentials on docking precision against cyclin-dependent kinase 2 protein: a multiple docking approach

The fundamental of molecular modeling is the interaction and binding to form a complex, because it explains the action of most drugs to a receptor active site. In the present study, different semiempirical (RM1, AM1, PM3, MNDO) and ab initio (HF, DFT) charge models were investigated for their performance in prediction of docking pose against CDK2 proteins with their respective inhibitor. Further, multiple docking approaches and Prime/MM-GBSA calculations were applied to predict the binding mode with respective charge model against CDK2 inhibitors. A reliable docking result was obtained using RRD, which showed significance improvement on ligand binding poses and docking score accuracy to the IFD. The combined use of RRD and Prime/MM-GBSA method could give a high correlation between the predicted binding free energy and experimental biological activity. The preliminary results point out that AM1 could be a precious charge model for design of new drugs with enhanced success rate. As a very similar result was also found for a different system of the protein-ligand binding, the suggested scoring function based on AM1 method seems to be applicable in drug design. The results from this study can provide insights into highest success rate for design of potent and selective CDK2 inhibitors.

Chemo- and bioinformatics resources for in silico drug discovery from medicinal plants beyond their traditional use: a critical review

An overview of databases andin silicotools for discovery of the hidden therapeutic potential of medicinal plants.

Receptor-ligand molecular docking

Docking methodology aims to predict the experimental binding modes and affinities of small molecules within the binding site of particular receptor targets and is currently used as a standard computational tool in drug design for lead compound optimisation and in virtual screening studies to find novel biologically active molecules. The basic tools of a docking methodology include a search algorithm and an energy scoring function for generating and evaluating ligand poses. In this review, we present the search algorithms and scoring functions most commonly used in current molecular docking methods that focus on protein-ligand applications. We summarise the main topics and recent computational and methodological advances in protein-ligand docking. Protein flexibility, multiple ligand binding modes and the free-energy landscape profile for binding affinity prediction are important and interconnected challenges to be overcome by further methodological developments in the docking field.

Structural properties of non-traditional drug targets present new challenges for virtual screening

Traditional drug targets have historically included signaling proteins that respond to small molecules and enzymes that use small molecules as substrates. Increasing attention is now being directed toward other types of protein targets, in particular those that exert their function by interacting with nucleic acids or other proteins rather than small-molecule ligands. Here, we systematically compare existing examples of inhibitors of protein-protein interactions to inhibitors of traditional drug targets. While both sets of inhibitors bind with similar potency, we find that the inhibitors of protein-protein interactions typically bury a smaller fraction of their surface area upon binding to their protein targets. The fact that an average atom is less buried suggests that more atoms are needed to achieve a given potency, explaining the observation that ligand efficiency is typically poor for inhibitors of protein-protein interactions. We then carried out a series of docking experiments and found a further consequence of these relatively exposed binding modes is that structure-based virtual screening may be more difficult: such binding modes do not provide sufficient clues to pick out active compounds from decoy compounds. Collectively, these results suggest that the challenges associated with such non-traditional drug targets may not lie with identifying compounds that potently bind to the target protein surface, but rather with identifying compounds that bind in a sufficiently buried manner to achieve good ligand efficiency and, thus, good oral bioavailability. While the number of available crystal structures of distinct protein interaction sites bound to small-molecule inhibitors is relatively small at present (only 21 such complexes were included in this study), these are sufficient to draw conclusions based on the current state of the field; as additional data accumulate it will be exciting to refine the viewpoint presented here. Even with this limited perspective however, we anticipate that these insights, together with new methods for exploring protein conformational fluctuations, may prove useful for identifying the "low-hanging fruit" among non-traditional targets for therapeutic intervention.

One Size Does Not Fit All: The Limits of Structure-Based Models in Drug Discovery

A major goal in computational chemistry has been to discover the set of rules that can accurately predict the binding affinity of any protein-drug complex, using only a single snapshot of its three-dimensional structure. Despite the continual development of structure-based models, predictive accuracy remains low, and the fundamental factors that inhibit the inference of all-encompassing rules have yet to be fully explored. Using statistical learning theory and information theory, here we prove that even the very best generalized structure-based model is inherently limited in its accuracy, and protein-specific models are always likely to be better. Our results refute the prevailing assumption that large data sets and advanced machine learning techniques will yield accurate, universally applicable models. We anticipate that the results will aid the development of more robust virtual screening strategies and scoring function error estimations.

Analysis of structure-based virtual screening studies and characterization of identified active compounds

Structure-based virtual screening makes explicit or implicit use of 3D target structure information to detect novel active compounds. Results of nearly 300 currently available original applications have been analyzed to characterize the state-of-the-art in this field. Compound selection from docking calculations is much influenced by subjective criteria. Although submicromolar compounds are identified, the majority of docking hits are only weakly potent. However, only a small percentage of docking hits can be reproduced by ligand-based methods. When docking calculations identify potent hits, they often originate from specialized compound sources (e.g., pharmaceutical compound decks or target-focused libraries) and also display a notable bias towards kinase targets. Structure-based virtual screening is the dominant approach to computational hit identification. Docking calculations frequently identify active compounds. Limited accuracy of compound scoring and ranking currently presents a major caveat of the approach that is often compensated for by chemical intuition and knowledge.

Exploring Structure-Activity Data Using the Landscape Paradigm

In this article we present an overview of the origin and applications of the activity landscape view of structure-actvitiy relationship data as conceived by Maggiora. Within this landscape, different regions exemplify different aspects of SAR trends - ranging from smoothly varying trends to discontinuous trends (also termed activity cliffs). We discuss the various definitions of landscapes and cliffs that have been proposed as well as different approaches to the numerical quantification of a landscape. We then highlight some of the landscape visualization approaches that have been developed, followed by a review of the various applications of activity landscapes and cliffs to topics in medicinal chemistry and SAR analysis.

Structure-based virtual screening for drug discovery: a problem-centric review

Structure-based virtual screening (SBVS) has been widely applied in early-stage drug discovery. From a problem-centric perspective, we reviewed the recent advances and applications in SBVS with a special focus on docking-based virtual screening. We emphasized the researchers' practical efforts in real projects by understanding the ligand-target binding interactions as a premise. We also highlighted the recent progress in developing target-biased scoring functions by optimizing current generic scoring functions toward certain target classes, as well as in developing novel ones by means of machine learning techniques.

Discovery of Novel α4β2 Neuronal Nicotinic Receptor Modulators through Structure-Based Virtual Screening

We performed a hierarchical structure-based virtual screening utilizing a comparative model of the human α4β2 neuronal nicotinic acetylcholine receptor (nAChR) extracellular domain. Compounds were selected for experimental testing based on structural diversity, binding pocket location, and standard error of the free energy scoring function used in the screening. Four of the eleven in silico hit compounds showed promising activity with low micromolar IC50 values in a calcium accumulation assay. Two of the antagonists were also proven to be selective for human α4β2 vs human α3β4 nAChRs. This is the first report of successful discovery of novel nAChR antagonists through the use of structure-based virtual screening with a human nAChR homology model. These compounds may serve as potential novel scaffolds for further development of selective nAChR antagonists.

Molecular modeling study of checkpoint kinase 1 inhibitors by multiple docking strategies and prime/MM-GBSA calculation

Developing chemicals that inhibit checkpoint kinase 1 (Chk1) is a promising adjuvant therapeutic to improve the efficacy and selectivity of DNA-targeting agents. Reliable prediction of binding-free energy and binding affinity of Chk1 inhibitors can provide a guide for rational drug design. In this study, multiple docking strategies and Prime/Molecular Mechanics Generalized Born Surface Area (Prime/MM-GBSA) calculation were applied to predict the binding mode and free energy for a series of benzoisoquinolinones as Chk1 inhibitors. Reliable docking results were obtained using induced-fit docking and quantum mechanics/molecular mechanics (QM/MM) docking, which showed superior performance on both ligand binding pose and docking score accuracy to the rigid-receptor docking. Then, the Prime/MM-GBSA method based on the docking complex was used to predict the binding-free energy. The combined use of QM/MM docking and Prime/MM-GBSA method could give a high correlation between the predicted binding-free energy and experimentally determined pIC(50) . The molecular docking combined with Prime/MM-GBSA simulation can not only be used to rapidly and accurately predict the binding-free energy of novel Chk1 inhibitors but also provide a novel strategy for lead discovery and optimization targeting Chk1.

The importance of ARG513 as a hydrogen bond anchor to discover COX-2 inhibitors in a virtual screening campaign

Structure-based virtual screening (SBVS) protocols were developed to find cyclooxygenase-2 (COX-2) inhibitors using the Protein-Ligand ANT System (PLANTS) docking software. The directory of useful decoys (DUD) dataset for COX-2 was used to retrospectively validate the protocols; the DUD consists of 426 known inhibitors in 13289 decoys. Based on criteria used in the article describing DUD datasets, the default protocol showed poor results. However, having ARG513 as a hydrogen bond anchor increased the quality of the SBVS protocol. The modified protocol showed results that could be well considered, with a maximum enrichment factor (EF(max)) value of 32.2.

Recent trends and observations in the design of high-quality screening collections

The design of a high-quality screening collection is of utmost importance for the early drug-discovery process and provides, in combination with high-quality assay systems, the foundation of future discoveries. Herein, we review recent trends and observations to successfully expand the access to bioactive chemical space, including the feedback from hit assessment interviews of high-throughput screening campaigns; recent successes with chemogenomics target family approaches, the identification of new relevant target/domain families, diversity-oriented synthesis and new emerging compound classes, and non-classical approaches, such as fragment-based screening and DNA-encoded chemical libraries. The role of in silico library design approaches are emphasized.

Knowledge-based scoring functions in drug design. 1. Developing a target-specific method for kinase-ligand interactions

Protein kinases are attractive targets for therapeutic interventions in many diseases. Due to their importance in drug discovery, a kinase family-specific potential of mean force (PMF) scoring function, kinase-PMF, was developed to assess the binding of ATP-competitive kinase inhibitors. It is hypothesized that target-specific PMF scoring functions may achieve increased performance in scoring along with the growth of the PDB database. The kinase-PMF inherits the functions and atom types in PMF04 and uses a kinase data set of 872 complexes to derive the potentials. The performance of kinase-PMF was evaluated with an external test set containing 128 kinase crystal structures. We compared it with eight scoring functions commonly used in computer-aided drug design, either in terms of the retrieval rate of retrieving "right" conformations or a virtual screening study. The evaluation results clearly demonstrate that a target-specific scoring function is a promising way to improve prediction power in structure-based drug design compared with other general scoring functions. To provide this rescoring service for researchers, a publicly accessible Web site was established at http://202.127.30.184:8080/scoring/index.jsp .

Computer-assisted combinatorial design of bicyclic thymidine analogs as inhibitors of Mycobacterium tuberculosis thymidine monophosphate kinase

Thymidine monophosphate kinase (TMPK(mt)) is an essential enzyme for nucleotide metabolism in Mycobacterium tuberculosis, and thus an attractive target for novel antituberculosis agents. In this work, we have explored the chemical space around the 2',3'-bicyclic thymidine nucleus by designing and in silico screening of a virtual focused library selected via structure based methods to identify more potent analogs endowed with favorable ADME-related properties. In all the library members we have exchanged the ribose ring of the template with a cyclopentane moiety that is less prone to enzymatic degradation. In addition, we have replaced the six-membered 2',3'-ring by a number of five-membered and six-membered heterocyclic rings containing alternative proton donor and acceptor groups, to exploit the interaction with the carboxylate groups of Asp9 and Asp163 as well as with several cationic residues present in the vicinity of the TMPK(mt) binding site. The three-dimensional structure of the TMPK(mt) complexed with 5-hydroxymethyl-dUMP, an analog of dTMP, was employed to develop a QSAR model, to parameterize a scoring function specific for the TMPK(mt) target and to select analogues which display the highest predicted binding to the target. As a result, we identified a small highly focused combinatorial subset of bicyclic thymidine analogues as virtual hits that are predicted to inhibit the mycobacterial TMPK in the submicromolar concentration range and to display favorable ADME-related properties.

Druggable pockets and binding site centric chemical space: a paradigm shift in drug discovery

Detection, comparison and analyses of binding pockets are pivotal to structure-based drug design endeavors, from hit identification, screening of exosites and de-orphanization of protein functions to the anticipation of specific and non-specific binding to off- and anti-targets. Here, we analyze protein-ligand complexes and discuss methods that assist binding site identification, prediction of druggability and binding site comparison. The full potential of pockets is yet to be harnessed, and we envision that better understanding of the pocket space will have far-reaching implications in the field of drug discovery, such as the design of pocket-specific compound libraries and scoring functions.

Frog2: Efficient 3D conformation ensemble generator for small compounds

Frog is a web tool dedicated to small compound 3D generation. Here we present the new version, Frog2, which allows the generation of conformation ensembles of small molecules starting from either 1D, 2D or 3D description of the compounds. From a compound description in one of the SMILES, SDF or mol2 formats, the server will return an ensemble of diverse conformers generated using a two stage Monte Carlo approach in the dihedral space. When starting from 1D or 2D description of compounds, Frog2 is capable to detect the sites of ambiguous stereoisomery, and thus to sample different stereoisomers. Frog2 also embeds new energy minimization and ring generation facilities that solve the problem of some missing cycle structures in the Frog1 ring library. Finally, the optimized generator of conformation ensembles in Frog2 results in a gain of computational time permitting Frog2 to be up to 20 times faster that Frog1, while producing satisfactory conformations in terms of structural quality and conformational diversity. The high speed and the good quality of generated conformational ensembles makes it possible the treatment of larger compound collections using Frog2. The server and documentation are freely available at http://bioserv.rpbs.univ-paris-diderot.fr/Frog2.

A knowledge-guided strategy for improving the accuracy of scoring functions in binding affinity prediction

Background

Current scoring functions are not very successful in protein-ligand binding affinity prediction albeit their popularity in structure-based drug designs. Here, we propose a general knowledge-guided scoring (KGS) strategy to tackle this problem. Our KGS strategy computes the binding constant of a given protein-ligand complex based on the known binding constant of an appropriate reference complex. A good training set that includes a sufficient number of protein-ligand complexes with known binding data needs to be supplied for finding the reference complex. The reference complex is required to share a similar pattern of key protein-ligand interactions to that of the complex of interest. Thus, some uncertain factors in protein-ligand binding may cancel out, resulting in a more accurate prediction of absolute binding constants.

Results

In our study, an automatic algorithm was developed for summarizing key protein-ligand interactions as a pharmacophore model and identifying the reference complex with a maximal similarity to the query complex. Our KGS strategy was evaluated in combination with two scoring functions (X-Score and PLP) on three test sets, containing 112 HIV protease complexes, 44 carbonic anhydrase complexes, and 73 trypsin complexes, respectively. Our results obtained on crystal structures as well as computer-generated docking poses indicated that application of the KGS strategy produced more accurate predictions especially when X-Score or PLP alone did not perform well.

Conclusions

Compared to other targeted scoring functions, our KGS strategy does not require any re-parameterization or modification on current scoring methods, and its application is not tied to certain systems. The effectiveness of our KGS strategy is in theory proportional to the ever-increasing knowledge of experimental protein-ligand binding data. Our KGS strategy may serve as a more practical remedy for current scoring functions to improve their accuracy in binding affinity prediction.