Normalizing molecular docking rankings using virtually generated decoys

Drug discovery research often relies on the use of virtual screening via molecular docking to identify active hits in compound libraries. An area for improvement among many state-of-the-art docking methods is the accuracy of the scoring functions used to differentiate active from nonactive ligands. Many contemporary scoring functions are influenced by the physical properties of the docked molecule. This bias can cause molecules with certain physical properties to incorrectly score better than others. Since variation in physical properties is inevitable in large screening libraries, it is desirable to account for this bias. In this paper, we present a method of normalizing docking scores using virtually generated decoy sets with matched physical properties. First, our method generates a set of property-matched decoys for every molecule in the screening library. Each library molecule and its decoy set are docked using a state-of-the-art method, producing a set of raw docking scores. Next, the raw docking score of each library molecule is normalized against the scores of its decoys. The normalized score represents the probability that the raw docking score was drawn from the background distribution of nonactive property-matched decoys. Assuming that the distribution of scores of active molecules differs from the nonactive score distribution, we expect that the score of an active compound will have a low probability of having been drawn from the nonactive score distribution. In addition to the use of decoys in normalizing docking scores, we suggest that decoy sets may be a useful tool to evaluate, improve, or develop scoring functions. We show that by analyzing docking scores of library molecules with respect to the docking scores of their virtually generated property-matched decoys, one can gain insight into the advantages, limitations, and reliability of scoring functions.     

OptiDock: virtual HTS of combinatorial libraries by efficient sampling of binding modes in product space

Products from combinatorial libraries generally share a common core structure that can be exploited to improve the efficiency of virtual high-throughput screening (vHTS). In general, it is more efficient to find a method that scales with the total number of reagents (Sigma growth) rather with the number of products (Pi growth). The OptiDock methodology described herein entails selecting a diverse but representative subset of compounds that span the structural space encompassed by the full library. These compounds are docked individually using the FlexX program (Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. J. Mol. Biol. 1995, 251, 470-489) to define distinct docking modes in terms of reference placements for combinatorial core atoms. Thereafter, substituents in R-cores (consisting of the core structure substituted at a single variation site) are docked, keeping the core atoms fixed at the coordinates dictated by each reference placement. Interaction energies are calculated for each docked R-core with respect to the target protein, and energies for whole compounds are calculated by finding the reference core placement for which the sum of corresponding R-core energies is most negative. The use of diverse whole compounds to define binding modes is a key advantage of the protocol over other combinatorial docking programs. As a result, OptiDock returns better-scoring conformers than does serially applied FlexX. OptiDock is also better able to find a viable docked pose for each library member than are other combinatorial approaches.     

High Speed Development and Synthesis of Novel Small Molecule Libraries

Combinatorial chemistry has produced libraries of millions of compounds in the last decade. Screening of those compounds, unfortunately, has not yet yielded as many new drug candidates as initially expected. Among a number of possible reasons, one is that many libraries combinatorial chemistry produced in the early periods are of the nature of linear, flat, and flexible molecules such as peptides and oligonucleotides, which do not have the desired properties to selectively interact with their targets to yield high quality hits and leads. In order to increase the number of quality hits and leads, rigid, structural featurerich and drug-like compound libraries are highly desirable. Design and development of structural features-rich and natural product-like combinatorial libraries, as well as high speed library production using modern solution and solid phase synthesis techniques such as IRORI's Directed Sorting technology, will be discussed.     

PLUMS: a program for the rapid optimization of focused libraries

PLUMS is a new method to perform rational monomer selection for combinatorial chemistry libraries. The algorithm has been developed to optimize focused libraries with specific two-dimensional and/or three-dimensional properties. A preliminary step is the identification of those molecules in the initial virtual library which satisfy the imposed property constraints; we define these molecules as the virtual hits. From the virtual hits, PLUMS generates a starting library, which is the true combinatorial library that includes all the virtual hits. Monomers are then removed in an iterative fashion, thus reducing the size of the library. At each iteration, the worst monomer is removed. Each sublibrary is selected using a global scoring function, which balances effectiveness and efficiency. The iterative process continues until one is left with a library that consists entirely of virtual hits. The optimal library, which is the best compromise between effectiveness and efficiency, can then be selected according to the score. During the iterative process, equivalent solutions may well occur and are taken into account by the algorithm, according to a user-defined parameter. The number of monomers for each substitution site and the size of the library are parameters that can be either optimized or used to constrain the selection. The results obtained on two test libraries are presented. PLUMS was compared with genetic algorithms (GA) and monomer frequency analysis (MFA), which are widely used for monomer selection. For the two test libraries, PLUMS and GA gave equivalent results. MFA is the fastest method, but it can give misleading solutions. Possible advantages and disadvantages of the different methods are discussed.     

Confirmation of primary active substances from high throughput screening of chemical and biological populations: a statistical approach and practical considerations

Many biologically important substances are discovered through screening of relevant chemical or biological libraries. The ability to find the active substances ("hits") from any random collection is largely determined by the quality of the assay and screening conditions. When a large population is screened for a specific characteristic, each member of that population is usually tested only once. Errors in the measurements require additional follow-up tests to confirm that each hit from the primary screen is truly active. In this report, we present a statistical model system that predicts the reliability of hits from a primary test as affected by the error in the assay and the choice of the hit threshold (hit limit). The hit confirmation rate, as well as false positive (representing substances that initially fall above the hit limit but whose true activity are below the hit limit) and false negative (representing substances that initially fall below the hit limit but whose true activity are in fact greater than the hit limit) rates have been analyzed with this model by computational simulation. This model can also be used in screen validation and post-screening data analysis. The statistical analysis presented here has broad implications and is applicable to screening of any large population for any specific characteristic. Obvious applications include drug discovery, gene chip analysis, population biology, directed molecular evolution, biological panning, and combinatorial material sciences.     

In Vitro Evaluation of In Silico Screening Approaches in Search for Selective ACE2 Binding Chemical Probes

New models for ACE2 receptor binding, based on QSAR and docking algorithms were developed, using XRD structural data and ChEMBL 26 database hits as training sets. The selectivity of the potential ACE2-binding ligands towards Neprilysin (NEP) and ACE was evaluated. The Enamine screening collection (3.2 million compounds) was virtually screened according to the above models, in order to find possible ACE2-chemical probes, useful for the study of SARS-CoV2-induced neurological disorders. An enzymology inhibition assay for ACE2 was optimized, and the combined diversified set of predicted selective ACE2-binding molecules from QSAR modeling, docking, and ultrafast docking was screened in vitro. The in vitro hits included two novel chemotypes suitable for further optimization.     

Virtual target screening: validation using kinase inhibitors

Computational methods involving virtual screening could potentially be employed to discover new biomolecular targets for an individual molecule of interest (MOI). However, existing scoring functions may not accurately differentiate proteins to which the MOI binds from a larger set of macromolecules in a protein structural database. An MOI will most likely have varying degrees of predicted binding affinities to many protein targets. However, correctly interpreting a docking score as a hit for the MOI docked to any individual protein can be problematic. In our method, which we term "Virtual Target Screening (VTS)", a set of small drug-like molecules are docked against each structure in the protein library to produce benchmark statistics. This calibration provides a reference for each protein so that hits can be identified for an MOI. VTS can then be used as tool for: drug repositioning (repurposing), specificity and toxicity testing, identifying potential metabolites, probing protein structures for allosteric sites, and testing focused libraries (collection of MOIs with similar chemotypes) for selectivity. To validate our VTS method, twenty kinase inhibitors were docked to a collection of calibrated protein structures. Here, we report our results where VTS predicted protein kinases as hits in preference to other proteins in our database. Concurrently, a graphical interface for VTS was developed.     

Combinatorial library design using a multiobjective genetic algorithm

Early results from screening combinatorial libraries have been disappointing with libraries either failing to deliver the improved hit rates that were expected or resulting in hits with characteristics that make them undesirable as lead compounds. Consequently, the focus in library design has shifted toward designing libraries that are optimized on multiple properties simultaneously, for example, diversity and "druglike" physicochemical properties. Here we describe the program MoSELECT that is based on a multiobjective genetic algorithm and which is able to suggest a family of solutions to multiobjective library design where all the solutions are equally valid and each represents a different compromise between the objectives. MoSELECT also allows the relationships between the different objectives to be explored with competing objectives easily identified. The library designer can then make an informed choice on which solution(s) to explore. Various performance characteristics of MoSELECT are reported based on a number of different combinatorial libraries.     

Virtual Screening and Structure Generation Applied to Drug Design

The methods of computer-aided drug design can be divided into two categories according to whether or not the structures of receptors are known1, corresponding to two principal strategies: (1) searching the bio-active ligands against virtual combinatorial libraries and calculating the affinity energy between ligand and receptor by docking ; (2) QSAR and 3D-structure data-mining. 3D-QSAR method is now applied widely to drug discovery, but this method is generally limited to refine the structu…     

Virtual screening of integrase inhibitors by large scale binding free energy calculations: the SAMPL4 challenge

As part of the SAMPL4 blind challenge, filtered AutoDock Vina ligand docking predictions and large scale binding energy distribution analysis method binding free energy calculations have been applied to the virtual screening of a focused library of candidate binders to the LEDGF site of the HIV integrase protein. The computational protocol leveraged docking and high level atomistic models to improve enrichment. The enrichment factor of our blind predictions ranked best among all of the computational submissions, and second best overall. This work represents to our knowledge the first example of the application of an all-atom physics-based binding free energy model to large scale virtual screening. A total of 285 parallel Hamiltonian replica exchange molecular dynamics absolute protein-ligand binding free energy simulations were conducted starting from docked poses. The setup of the simulations was fully automated, calculations were distributed on multiple computing resources and were completed in a 6-weeks period. The accuracy of the docked poses and the inclusion of intramolecular strain and entropic losses in the binding free energy estimates were the major factors behind the success of the method. Lack of sufficient time and computing resources to investigate additional protonation states of the ligands was a major cause of mispredictions. The experiment demonstrated the applicability of binding free energy modeling to improve hit rates in challenging virtual screening of focused ligand libraries during lead optimization.     

Accelerating high-throughput virtual screening through molecular pool-based active learning

Structure-based virtual screening is an important tool in early stage drug discovery that scores the interactions between a target protein and candidate ligands. As virtual libraries continue to grow (in excess of 10⁸ molecules), so too do the resources necessary to conduct exhaustive virtual screening campaigns on these libraries. However, Bayesian optimization techniques, previously employed in other scientific discovery problems, can aid in their exploration: a surrogate structure–property relationship model trained on the predicted affinities of a subset of the library can be applied to the remaining library members, allowing the least promising compounds to be excluded from evaluation. In this study, we explore the application of these techniques to computational docking datasets and assess the impact of surrogate model architecture, acquisition function, and acquisition batch size on optimization performance. We observe significant reductions in computational costs; for example, using a directed-message passing neural network we can identify 94.8% or 89.3% of the top-50 000 ligands in a 100M member library after testing only 2.4% of candidate ligands using an upper confidence bound or greedy acquisition strategy, respectively. Such model-guided searches mitigate the increasing computational costs of screening increasingly large virtual libraries and can accelerate high-throughput virtual screening campaigns with applications beyond docking.

Bayesian optimization can accelerate structure-based virtual screening campaigns by minimizing the total number of simulations performed while still identifying the vast majority of computational hits.     

Use of catalyst pharmacophore models for screening of large combinatorial libraries

Using a data set comprised of literature compounds and structure-activity data for cyclin dependent kinase 2, several pharmacophore hypotheses were generated using Catalyst and evaluated using several criteria. The two best were used in retrospective searches of 10 three-dimensional databases containing over 1,000,000 proprietary compounds. The results were then analyzed for the efficiency with which the hypotheses performed in the areas of compound prioritization, library prioritization, and library design. First as a test of their compound prioritization capabilities, the pharmacophore models were used to search combinatorial libraries that were known to contain CDK active compounds to see if the pharmacophore models could selectively choose the active compounds over the inactive compounds. Second as a test of their utility in library design again the pharmacophore models were used to search the active combinatorial libraries to see if the key synthons were over represented in the hits from the pharmacophore searches. Finally as a test of their ability to prioritize combinatorial libraries, several inactive libraries were searched in addition to the active libraries in order to see if the active libraries produced significantly more hits than the inactive libraries. For this study the pharmacophore models showed potential in all three areas. For compound prioritization, one of the models selected active compounds at a rate nearly 11 times that of random compound selection though in other cases models missed the active compounds entirely. For library design, most of the key fragments were over represented in the hits from at least one of the searches though again some key fragments were missed. Finally, for library prioritization, the two active libraries both produced a significant number of hits with both pharmacophore models, whereas none of the eight inactive libraries produced a significant number of hits for both models.     

Dihydropyrazole and dihydropyrrole structures based design of Kif15 inhibitors as novel therapeutic agents for cancer

Mitotic Kinesin motors, Eg5 and Kif15, have recently emerged as good targets for cancer as they play an inevitable role during mitosis. But, most of the Eg5 inhibitors were found ineffective when the cancer cells develop resistance to them by escalating the expression of Kif15 as alternative to Eg5. Therefore, the drugs that target Kif15 became necessary to be used either as a single or in combination with Eg5 inhibitors. The present study used 39 dihydropyrazole and 13 dihydropyrrole derivatives that were having in vitro inhibitory potential against kinesin motors to develop a common pharmacophore hypothesis AHRR and atom-based QSAR model. The model was used for virtual screening of ZINC database and the resultant hits were docked against Kif15. The four drug candidates with high docking score were examined for their activity and pharmacokinetic behaviour. Based on the results these drugs could be considered as lead candidates in further drug development for cancer.     

Development of a system to evaluate compound identity, purity, and concentration in a single experiment and its application in quality assessment of combinatorial libraries and screening hits

The development and use of a new assay system for the simultaneous determination of identity, purity, and concentration of sample components from combinatorial libraries produced by parallel synthesis are described. The system makes use of high-performance liquid chromatography with UV/vis photodiode array (PDA), evaporative light scattering (ELSD), chemiluminescent nitrogen (CLND), and time-of-flight mass spectrometer (TOFMS) detectors (HPLC-PDA-ELSD-CLND-TOFMS). Although these detectors have previously been utilized separately for the analysis of combinatorial chemistry libraries, the use of TOFMS along with CLND provides a synergistic combination enabling target and side-product structures to be identified and their concentrations and purities determined in a single experiment from a solution containing microgram levels of material. The CLND was found to give a linear response based on the number of moles of nitrogen present. Therefore, if the number of nitrogens per molecule is known, the concentration of each nitrogen-containing sample component may be determined utilizing an unrelated co-injected standard. A molecular formula for an impurity may often be calculated from the exact mass determined by the TOFMS and knowledge of the chemistry involved. Thus, if the sample components contain nitrogen, the concentration of every identified HPLC peak may be determined even in the absence of primary standards. This combination of detectors enabled the characterization of both target compounds and byproducts in combinatorial libraries, allowing the optimization of library synthetic procedures. This system was also used to survey the quality of libraries, enabling the selection of the best libraries for screening. This method also facilitated the characterization of samples from combinatorial libraries found as hits in high-throughput screening to establish the potency of the leads based on their actual concentration. In addition, concentrations and potencies of impurities were determined after identification of their structures, utilizing exact mass data, determination of charge states, and knowledge of the synthetic chemistry.     

Validation of an empirical RNA-ligand scoring function for fast flexible docking using RiboDock®

We report the design and validation of a fast empirical function for scoring RNA-ligand interactions, and describe its implementation within RiboDock, a virtual screening system for automated flexible docking. Building on well-known protein-ligand scoring function foundations, features were added to describe the interactions of common RNA-binding functional groups that were not handled adequately by conventional terms, to disfavour non-complementary polar contacts, and to control non-specific charged interactions. The results of validation experiments against known structures of RNA-ligand complexes compare favourably with previously reported methods. Binding modes were well predicted in most cases and good discrimination was achieved between native and non-native ligands for each binding site, and between native and non-native binding sites for each ligand. Further evidence of the ability of the method to identify true RNA binders is provided by compound selection ('enrichment factor') experiments based around a series of HIV-1 TAR RNA-binding ligands. Significant enrichment in true binders was achieved amongst high scoring docking hits, even when selection was from a library of structurally related, positively charged molecules. Coupled with a semi-automated cavity detection algorithm for identification of putative ligand binding sites, also described here, the method is suitable for the screening of very large databases of molecules against RNA and RNA-protein interfaces, such as those presented by the bacterial ribosome.