Molecular similarity based on DOCK-generated fingerprints

Computer-assisted drug discovery of potential natural inhibitors of the SARS-CoV-2 RNA-dependent RNA polymerase through a multi-phase in silico approach

BACKGROUND

The COVID-19 pandemic has led to significant loss of life and economic disruption worldwide. Currently, there are limited effective treatments available for this disease. SARS-CoV-2 RNA-dependent RNA polymerase (SARS-CoV-2 RdRp) has been identified as a potential target for drug development against COVID-19. Natural products have been shown to possess antiviral properties, making them a promising source for developing drugs against SARS-CoV-2.

OBJECTIVES

The objective of this study is to identify the most effective natural inhibitors of SARS-CoV-2 RdRp among a set of 4924 African natural products using a multi-phase in silico approach.

METHODS

The study utilized remdesivir (RTP), the co-crystallized ligand of RdRp, as a starting point to select compounds that have the most similar chemical structures among the examined set of compounds. Molecular fingerprints and structure similarity studies were carried out in the first part of the study. The second part of the study included molecular docking against SARS-CoV-2 RdRp (PDB ID: 7BV2) and Molecular Dynamics (MD) simulations including the calculation of RMSD, RMSF, Rg, SASA, hydrogen bonding, and PLIP. Moreover, the calculations of Molecular mechanics with generalised Born and surface area solvation (MM-GBSA) Lennard-Jones and Columbic electrostatic interaction energies have been conducted. Additionally, in silico ADMET and toxicity studies were performed to examine the drug likeness degrees of the selected compounds.

RESULTS

Eight compounds were identified as the most effective natural inhibitors of SARS-CoV-2 RdRp. These compounds are kaempferol 3-galactoside, kaempferol 3-O-β-D-glucopyranoside, mangiferin methyl ether, luteolin 7-O-β-D-glucopyranoside, quercetin-O-β-D-3-glucopyranoside, 1-methoxy-3-indolylmethyl glucosinolate, naringenin, and asphodelin A 4'-O-β-D-glucopyranoside.

CONCLUSION

The results of this study provide valuable information for the development of natural product-based drugs against COVID-19. However, the elected compounds should be further studied in vitro and in vivo to confirm their efficacy in treating COVID-19.

The Discovery of Potential SARS-CoV-2 Natural Inhibitors among 4924 African Metabolites Targeting the Papain-like Protease: A Multi-Phase In Silico Approach

Four compounds, hippacine, 4,2'-dihydroxy-4'-methoxychalcone, 2',5'-dihydroxy-4-methoxychalcone, and wighteone, were selected from 4924 African natural metabolites as potential inhibitors against SARS-CoV-2 papain-like protease (PLpro, PDB ID: 3E9S). A multi-phased in silico approach was employed to select the most similar metabolites to the co-crystallized ligand (TTT) of the PLpro through molecular fingerprints and structural similarity studies. Followingly, to examine the binding of the selected metabolites with the PLpro (molecular docking. Further, to confirm this binding through molecular dynamics simulations. Finally, in silico ADMET and toxicity studies were carried out to prefer the most convenient compounds and their drug-likeness. The obtained results could be a weapon in the battle against COVID-19 via more in vitro and in vivo studies.

Discovery of Potential SARS-CoV-2 Papain-like Protease Natural Inhibitors Employing a Multi-Phase In Silico Approach

As an extension of our research against COVID-19, a multiphase in silico approach was applied in the selection of the three most common inhibitors (Glycyrrhizoflavone (76), Arctigenin (94), and Thiangazole (298)) against papain-like protease, PLpro (PDB ID: 4OW0), among 310 metabolites of natural origin. All compounds of the exam set were reported as antivirals. The structural similarity between the examined compound set and S88, the co-crystallized ligand of PLpro, was examined through structural similarity and fingerprint studies. The two experiments pointed to Brevicollin (28), Cryptopleurine (41), Columbamine (46), Palmatine (47), Glycyrrhizoflavone (76), Licochalcone A (87), Arctigenin (94), Termilignan (98), Anolignan B (99), 4,5-dihydroxy-6″-deoxybromotopsentin (192), Dercitin (193), Tryptanthrin (200), 6-Cyano-5-methoxy-12-methylindolo [2, 3A] carbazole (211), Thiangazole (298), and Phenoxan (300). The binding ability against PLpro was screened through molecular docking, disclosing the favorable binding modes of six metabolites. ADMET studies expected molecules 28, 76, 94, 200, and 298 as the most favorable metabolites. Then, molecules 76, 94, and 298 were chosen through in silico toxicity studies. Finally, DFT studies were carried out on glycyrrhizoflavone (76) and indicated a high level of similarity in the molecular orbital analysis. The obtained data can be used in further in vitro and in vivo studies to examine and confirm the inhibitory effect of the filtered metabolites against PLpro and SARS-CoV-2.

Structure-Based Virtual Screening, Docking, ADMET, Molecular Dynamics, and MM-PBSA Calculations for the Discovery of Potential Natural SARS-CoV-2 Helicase Inhibitors from the Traditional Chinese Medicine

Continuing our antecedent work against COVID-19, a set of 5956 compounds of traditional Chinese medicine have been virtually screened for their potential against SARS-CoV-2 helicase (PDB ID: 5RMM). Initially, a fingerprint study with VXG, the ligand of the target enzyme, disclosed the similarity of 187 compounds. Then, a molecular similarity study declared the most similar 40 compounds. Subsequently, molecular docking studies were carried out to examine the binding modes and energies. Then, the most appropriate 26 compounds were subjected to in silico ADMET and toxicity studies to select the most convenient inhibitors to be: (1R,2S)-ephedrine (57), (1R,2S)-norephedrine (59), 2-(4-(pyrrolidin-1-yl)phenyl)acetic acid (84), 1-phenylpropane-1,2-dione (195), 2-methoxycinnamic acid (246), 2-methoxybenzoic acid (364), (R)-2-((R)-5-oxopyrrolidin-3-yl)-2-phenylacetic acid (405), (Z)-6-(3-hydroxy-4-methoxystyryl)-4-methoxy-2H-pyran-2-one (533), 8-chloro-2-(2-phenylethyl)-5,6,7-trihydroxy-5,6,7,8-tetrahydrochromone (637), 3-((1R,2S)-2-(dimethylamino)-1-hydroxypropyl)phenol (818), (R)-2-ethyl-4-(1-hydroxy-2-(methylamino)ethyl)phenol (5159), and (R)-2-((1S,2S,5S)-2-benzyl-5-hydroxy-4-methylcyclohex-3-en-1-yl)propane-1,2-diol (5168). Among the selected 12 compounds, the metabolites, compound 533 showed the best docking scores. Interestingly, the MD simulation studies for compound 533, the one with the highest docking score, over 100 ns showed its correct binding to SARS-CoV-2 helicase with low energy and optimum dynamics. Finally, MM-PBSA studies showed that 533 bonded favorably to SARS-CoV-2 helicase with a free energy value of −83 kJ/mol. Further, the free energy decomposition study determined the essential amino acid residues that contributed favorably to the binding process. The obtained results give a huge hope to find a cure for COVID-19 through further in vitro and in vivo studies for the selected compounds.

Ligand and Structure-Based In Silico Determination of the Most Promising SARS-CoV-2 nsp16-nsp10 2'-o-Methyltransferase Complex Inhibitors among 3009 FDA Approved Drugs

As a continuation of our earlier work against SARS-CoV-2, seven FDA-approved drugs were designated as the best SARS-CoV-2 nsp16-nsp10 2'-o-methyltransferase (2'OMTase) inhibitors through 3009 compounds. The in silico inhibitory potential of the examined compounds against SARS-CoV-2 nsp16-nsp10 2'-o-methyltransferase (PDB ID: (6W4H) was conducted through a multi-step screening approach. At the beginning, molecular fingerprints experiment with SAM (S-Adenosylmethionine), the co-crystallized ligand of the targeted enzyme, unveiled the resemblance of 147 drugs. Then, a structural similarity experiment recommended 26 compounds. Therefore, the 26 compounds were docked against 2'OMTase to reveal the potential inhibitory effect of seven promising compounds (Protirelin, (1187), Calcium folinate (1913), Raltegravir (1995), Regadenoson (2176), Ertapenem (2396), Methylergometrine (2532), and Thiamine pyrophosphate hydrochloride (2612)). Out of the docked ligands, Ertapenem (2396) showed an ideal binding mode like that of the co-crystallized ligand (SAM). It occupied all sub-pockets of the active site and bound the crucial amino acids. Accordingly, some MD simulation experiments (RMSD, RMSF, Rg, SASA, and H-bonding) have been conducted for the 2'OMTase-Ertapenem complex over 100 ns. The performed MD experiments verified the correct binding mode of Ertapenem against 2'OMTase exhibiting low energy and optimal dynamics. Finally, MM-PBSA studies indicated that Ertapenem bonded advantageously to the targeted protein with a free energy value of -43 KJ/mol. Furthermore, the binding free energy analysis revealed the essential amino acids of 2'OMTase that served positively to the binding. The achieved results bring hope to find a treatment for COVID-19 via in vitro and in vivo studies for the pointed compounds.

Multi-Phase In Silico Discovery of Potential SARS-CoV-2 RNA-Dependent RNA Polymerase Inhibitors among 3009 Clinical and FDA-Approved Related Drugs

Proceeding our prior studies of SARS-CoV-2, the inhibitory potential against SARS-CoV-2 RNA-dependent RNA polymerase (RdRp) has been investigated for a collection of 3009 clinical and FDA-approved drugs. A multi-phase in silico approach has been employed in this study. Initially, a molecular fingerprint experiment of Remdesivir (RTP), the co-crystallized ligand of the examined protein, revealed the most similar 150 compounds. Among them, 30 compounds were selected after a structure similarity experiment. Subsequently, the most similar 30 compounds were docked against SARS-CoV-2 RNA-dependent RNA polymerase (PDB ID: 7BV2). Aloin 359, Baicalin 456, Cefadroxil 1273, Sophoricoside 1459, Hyperoside 2109, and Vitexin 2286 exhibited the most precise binding modes, as well as the best binding energies. To confirm the obtained results, MD simulations experiments have been conducted for Hyperoside 2109, the natural flavonoid glycoside that exhibited the best docking scores, against RdRp (PDB ID: 7BV2) for 100 ns. The achieved results authenticated the correct binding of 2109, showing low energy and optimum dynamics. Our team presents these outcomes for scientists all over the world to advance in vitro and in vivo examinations against COVID-19 for the promising compounds.

QSAR-derived affinity fingerprints (part 1): fingerprint construction and modeling performance for similarity searching, bioactivity classification and scaffold hopping

An affinity fingerprint is the vector consisting of compound’s affinity or potency against the reference panel of protein targets. Here, we present the QAFFP fingerprint, 440 elements long in silico QSAR-based affinity fingerprint, components of which are predicted by Random Forest regression models trained on bioactivity data from the ChEMBL database. Both real-valued (rv-QAFFP) and binary (b-QAFFP) versions of the QAFFP fingerprint were implemented and their performance in similarity searching, biological activity classification and scaffold hopping was assessed and compared to that of the 1024 bits long Morgan2 fingerprint (the RDKit implementation of the ECFP4 fingerprint). In both similarity searching and biological activity classification, the QAFFP fingerprint yields retrieval rates, measured by AUC (~ 0.65 and ~ 0.70 for similarity searching depending on data sets, and ~ 0.85 for classification) and EF5 (~ 4.67 and ~ 5.82 for similarity searching depending on data sets, and ~ 2.10 for classification), comparable to that of the Morgan2 fingerprint (similarity searching AUC of ~ 0.57 and ~ 0.66, and EF5 of ~ 4.09 and ~ 6.41, depending on data sets, classification AUC of ~ 0.87, and EF5 of ~ 2.16). However, the QAFFP fingerprint outperforms the Morgan2 fingerprint in scaffold hopping as it is able to retrieve 1146 out of existing 1749 scaffolds, while the Morgan2 fingerprint reveals only 864 scaffolds.

Bioturbo similarity searching: combining chemical and biological similarity to discover structurally diverse bioactive molecules

Virtual screening using bioactivity profiles has become an integral part of currently applied hit finding methods in pharmaceutical industry. However, a significant drawback of this approach is that it is only applicable to compounds that have been biologically tested in the past and have sufficient activity annotations for meaningful profile comparisons. Although bioactivity data generated in pharmaceutical institutions are growing on an unprecedented scale, the number of biologically annotated compounds still covers only a minuscule fraction of chemical space. For a newly synthesized compound or an isolated natural product to be biologically characterized across multiple assays, it may take a considerable amount of time. Consequently, this chemical matter will not be included in virtual screening campaigns based on bioactivity profiles. To overcome this problem, we herein introduce bioturbo similarity searching that uses chemical similarity to map molecules without biological annotations into bioactivity space and then searches for biologically similar compounds in this reference system. In benchmark calculations on primary screening data, we demonstrate that our approach generally achieves higher hit rates and identifies structurally more diverse compounds than approaches using chemical information only. Furthermore, our method is able to discover hits with novel modes of inhibition that traditional 2D and 3D similarity approaches are unlikely to discover. Test calculations on a set of natural products reveal the practical utility of the approach for identifying novel and synthetically more accessible chemical matter.

Virtual affinity fingerprints for target fishing: a new application of Drug Profile Matching

We recently introduced Drug Profile Matching (DPM), a novel virtual affinity fingerprinting bioactivity prediction method. DPM is based on the docking profiles of ca. 1200 FDA-approved small-molecule drugs against a set of nontarget proteins and creates bioactivity predictions based on this pattern. The effectiveness of this approach was previously demonstrated for therapeutic effect prediction of drug molecules. In the current work, we investigated the applicability of DPM for target fishing, i.e. for the prediction of biological targets for compounds. Predictions were made for 77 targets, and their accuracy was measured by Receiver Operating Characteristic (ROC) analysis. Robustness was tested by a rigorous 10-fold cross-validation procedure. This procedure identified targets (N = 45) with high reliability based on DPM performance. These 45 categories were used in a subsequent study which aimed at predicting the off-target profiles of currently approved FDA drugs. In this data set, 79% of the known drug-target interactions were correctly predicted by DPM, and additionally 1074 new drug-target interactions were suggested. We focused our further investigation on the suggested interactions of antipsychotic molecules and confirmed several interactions by a review of the literature.

SAGPAR: structural grammar-based automated pathway reconstruction

In-silico metabolic engineering is a very useful branch of systems biology for modeling, analysis and prediction of various outcomes of metabolic pathways. It can also be used for detecting interactions and dynamics within a network. Various protocols have been proposed for modeling a pathway. But most of these protocols have various disadvantages and shortcomings with respect to automated pathway modeling and analysis. In the present article, we have proposed a novel algorithm for automated pathway reconstruction. We have also made a comparative study of our algorithm with other standard protocols and discussed its advantages over others. We present StructurAl Grammar-based automated PAthway Reconstruction (SAGPAR), a fast and robust algorithm that generates any metabolic pathway using some given structural representations of metabolites. Users can model any pathway based on some pre-required features that are asked as an input by the algorithm. The algorithm also takes into considerations various thermodynamic thresholds and structural properties while modeling a pathway. The given algorithm has been tested on the standard pathway datasets of 25 pathways of Mycoplasma pneumoniae M129 and 24 pathways of Homo sapiens. The dataset is taken from KEGG and PubChem Compound data repositories. SAGPAR performs much better than some already present metabolic pathway analysis tools like Copasi, PHT, Gepasi, Jarnac and Path-A.

Drug effect prediction by polypharmacology-based interaction profiling

Most drugs exert their effects via multitarget interactions, as hypothesized by polypharmacology. While these multitarget interactions are responsible for the clinical effect profiles of drugs, current methods have failed to uncover the complex relationships between them. Here, we introduce an approach which is able to relate complex drug-protein interaction profiles with effect profiles. Structural data and registered effect profiles of all small-molecule drugs were collected, and interactions to a series of nontarget protein binding sites of each drug were calculated. Statistical analyses confirmed a close relationship between the studied 177 major effect categories and interaction profiles of ca. 1200 FDA-approved small-molecule drugs. On the basis of this relationship, the effect profiles of drugs were revealed in their entirety, and hitherto uncovered effects could be predicted in a systematic manner. Our results show that the prediction power is independent of the composition of the protein set used for interaction profile generation.

Extended-connectivity fingerprints

Extended-connectivity fingerprints (ECFPs) are a novel class of topological fingerprints for molecular characterization. Historically, topological fingerprints were developed for substructure and similarity searching. ECFPs were developed specifically for structure-activity modeling. ECFPs are circular fingerprints with a number of useful qualities: they can be very rapidly calculated; they are not predefined and can represent an essentially infinite number of different molecular features (including stereochemical information); their features represent the presence of particular substructures, allowing easier interpretation of analysis results; and the ECFP algorithm can be tailored to generate different types of circular fingerprints, optimized for different uses. While the use of ECFPs has been widely adopted and validated, a description of their implementation has not previously been presented in the literature.

Analysis and comparison of 2D fingerprints: insights into database screening performance using eight fingerprint methods

Virtual screening is a widely used strategy in modern drug discovery and 2D fingerprint similarity is an important tool that has been successfully applied to retrieve active compounds from large datasets. However, it is not always straightforward to select an appropriate fingerprint method and associated settings for a given problem. Here, we applied eight different fingerprint methods, as implemented in the new cheminformatics package Canvas, on a well-validated dataset covering five targets. The fingerprint methods include Linear, Dendritic, Radial, MACCS, MOLPRINT2D, Pairwise, Triplet, and Torsion. We find that most fingerprints have similar retrieval rates on average; however, each has special characteristics that distinguish its performance on different query molecules and ligand sets. For example, some fingerprints exhibit a significant ligand size dependency whereas others are more robust with respect to variations in the query or active compounds. In cases where little information is known about the active ligands, MOLPRINT2D fingerprints produce the highest average retrieval actives. When multiple queries are available, we find that a fingerprint averaged over all query molecules is generally superior to fingerprints derived from single queries. Finally, a complementarity metric is proposed to determine which fingerprint methods can be combined to improve screening results.

A similarity search using molecular topological graphs

A molecular similarity measure has been developed using molecular topological graphs and atomic partial charges. Two kinds of topological graphs were used. One is the ordinary adjacency matrix and the other is a matrix which represents the minimum path length between two atoms of the molecule. The ordinary adjacency matrix is suitable to compare the local structures of molecules such as functional groups, and the other matrix is suitable to compare the global structures of molecules. The combination of these two matrices gave a similarity measure. This method was applied to in silico drug screening, and the results showed that it was effective as a similarity measure.

A new method for in-silico drug screening and similarity search using molecular dynamics maximum volume overlap (MD-MVO) method

We developed a new molecular dynamics simulation method for molecular overlapping (alignment) and ligand-based in-silico drug screening based on molecular similarity. The molecular system consists of the query compound and the other compound(s) selected from a compound library. The newly introduced intermolecular interaction between compounds is proportional to the molecular overlap instead of the van der Waals and Coulomb interactions between atoms of different molecules. This method was able to achieve both conformer generation of molecules and molecular overlapping (alignment) at the same time. After an energy minimization and following short-time MD simulation, the molecules in the system were overlapped with each other and the similarity between compounds was measured by the volume of the overlap. We applied this MD simulation method to ligand-based in-silico drug screening and found that it worked well for several targets.

In-silico drug screening method based on the protein–compound affinity matrix using the factor selection technique

We have developed a new in-silico drug screening method, a modified version of a docking score index (DSI) method, based on a protein-compound docking affinity matrix. By using this method, the docking scores are converted to the docking score indexes by the principal component analysis (PCA) method and each compound is projected into a PCA space. In this study, we propose a method to select a set of suitable principal component axes and evaluate the database enrichment for 12 target proteins. This method selects the new active compounds or hits, which are close to the known active compounds, thereby enhancing the database enrichment.

Finding ligands for G protein-coupled receptors based on the protein–compound affinity matrix

We developed a novel method of identifying new active ligands based on information related to known active compounds using protein-compound docking simulations, even when the tertiary structure of the actual target receptor protein is unknown. This method was used to find ligands of G protein-coupled receptors (GPCRs), i.e., agonists and antagonists of histamine, adrenaline, serotonin and dopamine receptors. The principal component analysis (PCA) method was applied to the protein-compound affinity matrix, which was given by thorough docking calculations between sets of many protein pockets and chemical compounds. The set of protein pockets did not necessary include the target protein. Each compound was depicted as a point in the PCA space. Compounds in a sphere, whose center was set to the known active compound in the multi-dimensional PCA space or to the average position of several known active compounds, were selected as candidate-hit compounds. Our method was found to be effective for finding the ligands of GPCRs based on known native ligands, even when only the soluble protein structures were used in the docking simulations.

An Efficient in Silico Screening Method Based on the Protein−Compound Affinity Matrix and Its Application to the Design of a Focused Library for Cytochrome P450 (CYP) Ligands

A new method has been developed to design a focused library based on available active compounds using protein-compound docking simulations. This method was applied to the design of a focused library for cytochrome P450 (CYP) ligands, not only to distinguish CYP ligands from other compounds but also to identify the putative ligands for a particular CYP. Principal component analysis (PCA) was applied to the protein-compound affinity matrix, which was obtained by thorough docking calculations between a large set of protein pockets and chemical compounds. Each compound was depicted as a point in the PCA space. Compounds that were close to the known active compounds were selected as candidate hit compounds. A machine-learning technique optimized the docking scores of the protein-compound affinity matrix to maximize the database enrichment of the known active compounds, providing an optimized focused library.

Reasoning about molecular similarity and properties

Ascertaining the similarity amongst molecules is a fundamental problem in biology and drug discovery. Since similar molecules tend to have similar biological properties, the notion of molecular similarity plays an important role in exploration of molecular structural space, query-retrieval in molecular databases, and in structure-activity modeling. This problem is related to the issue of molecular representation. Currently, approaches with high descriptive power like 3D surface-based representations are available. However, most techniques tend to focus on 2D graph-based molecular similarity due to the complexity that accompanies reasoning with more elaborate representations. This paper addresses the problem of determining similarity when molecules are described using complex surface-based representations. It proposes an intrinsic, spherical representation that systematically maps points on a molecular surface to points on a standard coordinate system (a sphere). Molecular geometry, molecular fields, and effects due to field super-positioning can then be captured as distributions on the surface of the sphere. Molecular similarity is obtained by computing the similarity of the corresponding property distributions using a novel formulation of histogram-intersection. This method is robust to noise, obviates molecular pose-optimization, can incorporate conformational variations, and facilitates highly efficient determination of similarity. Retrieval performance, applications in structure-activity modeling of complex biological properties, and comparisons with existing research and commercial methods demonstrate the validity and effectiveness of the approach.

Robust ligand-based modeling of the biological targets of known drugs

Systematic annotation of the primary targets of roughly 1000 known therapeutics reveals that over 700 of these modulate approximately 85 biological targets. We report the results of three analyses. In the first analysis, drug/drug similarities and target/target similarities were computed on the basis of three-dimensional ligand structures. Drug pairs sharing a target had significantly higher similarity than drug pairs sharing no target. Also, target pairs with no overlap in annotated drug specificity shared lower similarity than target pairs with increasing overlap. Two-way agglomerative clusterings of drugs and targets were consistent with known pharmacology and suggestive that side effects and drug-drug interactions might be revealed by modeling many targets. In the second analysis, we constructed and tested ligand-based models of 22 diverse targets in virtual screens using a background of screening molecules. Greater than 100-fold enrichment of cognate versus random molecules was observed in 20/22 cases. In the third analysis, selectivity of the models was tested using a background of drug molecules, with selectivity of greater than 80-fold observed in 17/22 cases. Predicted activities derived from crossing drugs against modeled targets identified a number of known side effects, drug specificities, and drug-drug interactions that have a rational basis in molecular structure.

Probing drug action using in vitro pharmacological profiles

The history of structure-activity relationships in drug design represents a long search for appropriate descriptors of broad biological action at the molecular level. In this context, recent work showing that in vitro pharmacological profiles can be used as exquisite descriptors of the broad biological effects of compounds represents an important breakthrough. Generalization of the methodology could have important implications for drug discovery and development. It might also provide a novel and insightful way to study systems biology.

A Discussion of Measures of Enrichment in Virtual Screening: Comparing the Information Content of Descriptors with Increasing Levels of Sophistication

We have performed virtual screening using some very simple features, by employing the number of atoms per element as molecular descriptors but without regard to any structural information whatsoever. Surprisingly, these atom counts are able to outperform virtual-affinity-based fingerprints and Unity fingerprints in some activity classes. Although molecular weight and other biases were known in target-based virtual screening settings (docking), we report the effect of using very simple descriptors for ligand-based virtual screening, by using clearly defined biological targets and employing a large data set (>100,000 compounds) containing multiple (11) activity classes. Structure-unaware atom count vectors as descriptors in combination with the Euclidean distance measure are able to achieve "enrichment factors" over random selection of around 4 (depending on the particular class of active compounds), putting the enrichment factors reported for more sophisticated virtual screening methods in a different light. They are also able to retrieve active compounds with novel scaffolds instead of merely the expected structural analogues. The added value of many currently used virtual screening methods (calculated as enrichment factors) drops down to a factor of between 1 and 2, instead of often reported double-digit figures. The observed effect is much less profound for simple descriptors such as molecular weight and is only present in cases of atypical (larger) ligands. The current state of virtual screening is not as sophisticated as might be expected, which is due to descriptors still not being able to capture structural properties relevant to binding. This fact can partly be explained by highly nonlinear structure-activity relationships, which represent a severe limitation of the "similar property principle" in the context of bioactivity.

Molecular surface point environments for virtual screening and the elucidation of binding patterns (MOLPRINT 3D)

A novel method (MOLPRINT 3D) for virtual screening and the elucidation of ligand-receptor binding patterns is introduced that is based on environments of molecular surface points. The descriptor uses points relative to the molecular coordinates, thus it is translationally and rotationally invariant. Due to its local nature, conformational variations cause only minor changes in the descriptor. If surface point environments are combined with the Tanimoto coefficient and applied to virtual screening, they achieve retrieval rates comparable to that of two-dimensional (2D) fingerprints. The identification of active structures with minimal 2D similarity ("scaffold hopping") is facilitated. In combination with information-gain-based feature selection and a naive Bayesian classifier, information from multiple molecules can be combined and classification performance can be improved. Selected features are consistent with experimentally determined binding patterns. Examples are given for angiotensin-converting enzyme inhibitors, 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors, and thromboxane A2 antagonists.

Docking and scoring in virtual screening for drug discovery: methods and applications

Computational approaches that 'dock' small molecules into the structures of macromolecular targets and 'score' their potential complementarity to binding sites are widely used in hit identification and lead optimization. Indeed, there are now a number of drugs whose development was heavily influenced by or based on structure-based design and screening strategies, such as HIV protease inhibitors. Nevertheless, there remain significant challenges in the application of these approaches, in particular in relation to current scoring schemes. Here, we review key concepts and specific features of small-molecule-protein docking methods, highlight selected applications and discuss recent advances that aim to address the acknowledged limitations of established approaches.

Protein classification using comparative molecular interaction profile analysis system

We recently introduced a new molecular description factor, interaction profile Factor (IPF) that is useful for evaluating molecular interactions. IPF is a data set of interaction energies calculated by the Comparative Molecular Interaction Profile Analysis system (CoMIPA). CoMIPA utilizes AutoDock 3.0 docking program, and the system has shown to be a powerful tool in clustering the interacting properties between small molecules and proteins. In this report, we describe the application of CoMIPA for protein clustering. A sample set of 15 proteins that share less than 20% homology and have no common functional motifs in primary structure were chosen. Using CoMIPA, we were able to cluster proteins that bound to the same small molecule. Other structural homology-based clustering programs such as PSI-BLAST or PFAM were unable to achieve the same classification. The results are striking because it is difficult to find any common features in the active sites of these proteins that share the same ligand. CoMIPA adds new dimensions for protein classification and has the potential to be a helpful tool in predicting and analyzing molecular interactions.

Similarity Searching of Chemical Databases Using Atom Environment Descriptors (MOLPRINT 2D): Evaluation of Performance

A molecular similarity searching technique based on atom environments, information-gain-based feature selection, and the naive Bayesian classifier has been applied to a series of diverse datasets and its performance compared to those of alternative searching methods. Atom environments are count vectors of heavy atoms present at a topological distance from each heavy atom of a molecular structure. In this application, using a recently published dataset of more than 100000 molecules from the MDL Drug Data Report database, the atom environment approach appears to outperform fusion of ranking scores as well as binary kernel discrimination, which are both used in combination with Unity fingerprints. Overall retrieval rates among the top 5% of the sorted library are nearly 10% better (more than 14% better in relative numbers) than those of the second best method, Unity fingerprints and binary kernel discrimination. In 10 out of 11 sets of active compounds the combination of atom environments and the naive Bayesian classifier appears to be the superior method, while in the remaining dataset, data fusion and binary kernel discrimination in combination with Unity fingerprints is the method of choice. Binary kernel discrimination in combination with Unity fingerprints generally comes second in performance overall. The difference in performance can largely be attributed to the different molecular descriptors used. Atom environments outperform Unity fingerprints by a large margin if the combination of these descriptors with the Tanimoto coefficient is compared. The naive Bayesian classifier in combination with information-gain-based feature selection and selection of a sensible number of features performs about as well as binary kernel discrimination in experiments where these classification methods are compared. When used on a monoaminooxidase dataset, atom environments and the naive Bayesian classifier perform as well as binary kernel discrimination in the case of a 50/50 split of training and test compounds. In the case of sparse training data, binary kernel discrimination is found to be superior on this particular dataset. On a third dataset, the atom environment descriptor shows higher retrieval rates than other 2D fingerprints tested here when used in combination with the Tanimoto similarity coefficient. Feature selection is shown to be a crucial step in determining the performance of the algorithm. The representation of molecules by atom environments is found to be more effective than Unity fingerprints for the type of biological receptor similarity calculations examined here. Combining information prior to scoring and including information about inactive compounds, as in the Bayesian classifier and binary kernel discrimination, is found to be superior to posterior data fusion (in the datasets tested here).

Molecular Similarity Searching Using Atom Environments, Information-Based Feature Selection, and a Naïve Bayesian Classifier

A novel technique for similarity searching is introduced. Molecules are represented by atom environments, which are fed into an information-gain-based feature selection. A naïve Bayesian classifier is then employed for compound classification. The new method is tested by its ability to retrieve five sets of active molecules seeded in the MDL Drug Data Report (MDDR). In comparison experiments, the algorithm outperforms all current retrieval methods assessed here using two- and three-dimensional descriptors and offers insight into the significance of structural components for binding.

Similarity searching using reduced graphs

Reduced graphs provide summary representations of chemical structures. In this work, the effectiveness of reduced graphs for similarity searching is investigated. Different types of reduced graphs are introduced that aim to summarize features of structures that have the potential to form interactions with receptors while retaining the topology between the features. Similarity searches have been carried out across a variety of different activity classes. The effectiveness of the reduced graphs at retrieving compounds with the same activity as known target compounds is compared with searching using Daylight fingerprints. The reduced graphs are shown to be effective for similarity searching and to retrieve more diverse active compounds than those found using Daylight fingerprints; they thus represent a complementary similarity searching tool.

Flexsim-R: a virtual affinity fingerprint descriptor to calculate similarities of functional groups

Methods to describe the similarity of fragments occurring in drug-like molecules are of fundamental importance in computational drug design. In the early phase of lead discovery, they can help to select diverse building blocks for combinatorial compound libraries intended for broad screening. In lead optimization, such methods can guide bioisosteric replacements of one functional group by another or serve as descriptors for QSAR calculations. In this paper, we outline the development of a novel 3D descriptor, termed Flexsim-R, which is a further extension of our virtual affinity fingerprint idea. Descriptors are calculated based on docking of small fragments such as building blocks for combinatorial chemistry or functional groups of drug-like molecules into a reference panel of protein binding sites. The method is validated by examining the neighborhood behavior of the affinity fingerprints and by deriving predictive QSAR models for a couple of literature peptide data sets.

Reoptimization of MDL keys for use in drug discovery

For a number of years MDL products have exposed both 166 bit and 960 bit keysets based on 2D descriptors. These keysets were originally constructed and optimized for substructure searching. We report on improvements in the performance of MDL keysets which are reoptimized for use in molecular similarity. Classification performance for a test data set of 957 compounds was increased from 0.65 for the 166 bit keyset and 0.67 for the 960 bit keyset to 0.71 for a surprisal S/N pruned keyset containing 208 bits and 0.71 for a genetic algorithm optimized keyset containing 548 bits. We present an overview of the underlying technology supporting the definition of descriptors and the encoding of these descriptors into keysets. This technology allows definition of descriptors as combinations of atom properties, bond properties, and atomic neighborhoods at various topological separations as well as supporting a number of custom descriptors. These descriptors can then be used to set one or more bits in a keyset. We constructed various keysets and optimized their performance in clustering bioactive substances. Performance was measured using methodology developed by Briem and Lessel. "Directed pruning" was carried out by eliminating bits from the keysets on the basis of random selection, values of the surprisal of the bit, or values of the surprisal S/N ratio of the bit. The random pruning experiment highlighted the insensitivity of keyset performance for keyset lengths of more than 1000 bits. Contrary to initial expectations, pruning on the basis of the surprisal values of the various bits resulted in keysets which underperformed those resulting from random pruning. In contrast, pruning on the basis of the surprisal S/N ratio was found to yield keysets which performed better than those resulting from random pruning. We also explored the use of genetic algorithms in the selection of optimal keysets. Once more the performance was only a weak function of keyset size, and the optimizations failed to identify a single globally optimal keyset. Instead multiple, equally optimal keysets could be produced which had relatively low overlap of the descriptors they encoded.

Do structurally similar molecules have similar biological activity?

To design diverse combinatorial libraries or to select diverse compounds to augment a screening collection, computational chemists frequently reject compounds that are > or =0.85 similar to one already chosen for the combinatorial library or in the screening set. Using Daylight fingerprints, this report shows that for IC(50) values determined as a follow-up to 115 high-throughput screening assays, there is only a 30% chance that a compound that is > or = 0.85 (Tanimoto) similar to an active is itself active. Although this enrichment is greater than that found with random screening and docking to three-dimensional structures, this low fraction of actives within similar compounds occurs not only because of deficiencies in the Daylight fingerprints and Tanimoto similarity calculations but also because similar compounds do not necessarily interact with the target macromolecule in similar ways. The current study emphasizes the statistical or probabilistic nature of library design and that perfect results cannot be expected.

Comments on the design of chemical libraries for screening

Different representations of molecules, based on distinct sets of properties can yield different perspectives of the issues involved in library design. In particular, different chemical representations can give rise to very different estimates of required library sizes. We provide a preliminary mathematical framework that examines the size of libraries required to adequately sample the spaces corresponding to some commonly used property sets. Introduction of conformational flexibility is also discussed as a means of increasing coverage of chemical libraries, while at the same time considering the thermodynamic consequences of flexibility upon detectable activity. Our theoretical analysis reveals that the property spaces currently in use are extremely large and unlikely to provide adequate discrimination among compounds.

Measuring molecular similarity and diversity: total pharmacophore diversity

A novel method, total pharmacophore diversity (ToPD), based on known pharmacophore features for numerically defining molecular similarity or diversity is described. The method captures the 3D shape and functionality of molecules by the analysis of relevant intramolecular distances to generate a short and descriptive pharmacophoric fingerprint for each molecule. The ToPD fingerprints can then be used in diversity analysis, clustering, or database searching. Conformational sampling is carried out when needed by the means of molecular dynamics. Our results show that ToPD outperforms a traditional 2D fingerprint technique in all test cases.

Anti-HIV activity of medicinal herbs: usage and potential development

The acquired immunodeficiency syndrome (AIDS) is a result of human immunodeficiency virus (HIV) infection which subsequently leads to significant suppression of immune functions. AIDS is a significant threat to the health of mankind, and the search for effective therapies to treat AIDS is of paramount importance. Several chemical anti-HIV agents have been developed. However, besides the high cost, there are adverse effects and limitations associated with using chemotherapy for the treatment of HIV infection. Thus, herbal medicines have frequently been used as an alternative medical therapy by HIV positive individuals and AIDS patients. The aim of this review is to summarize research findings for herbal medicines, which are endowed with the ability to inhibit HIV. In this article, we will emphasize a Chinese herbal medicine, Scutellaria baicalensis Georgi and its identified components (i.e., baicalein and baicalin), which have been shown to inhibit infectivity and replication of HIV. Potential development of anti-AIDS compounds using molecular modeling methods will also be discussed.

Design, docking, and evaluation of multiple libraries against multiple targets

We present a general approach to the design, docking, and virtual screening of multiple combinatorial libraries against a family of proteins. The method consists of three main stages: docking the scaffold, selecting the best substituents at each site of diversity, and comparing the resultant molecules within and between the libraries. The core "divide-and-conquer" algorithm for side-chain selection, developed from an earlier version (Sun et al., J Comp Aided Mol Design 1998;12:597-604), provides a way to explore large lists of substituents with linear rather than combinatorial time dependence. We have applied our method to three combinatorial libraries and three serine proteases: trypsin, chymotrypsin, and elastase. We show that the scaffold docking procedure, in conjunction with a novel vector-based orientation filter, reproduces crystallographic binding modes. In addition, the free-energy-based scoring procedure (Zou et al., J Am Chem Soc 1999;121:8033-8043) is able to reproduce experimental binding data for P1 mutants of macromolecular protease inhibitors. Finally, we show that our method discriminates between a peptide library and virtual libraries built on benzodiazepine and tetrahydroisoquinolinone scaffolds. Implications of the docking results for library design are explored.