Ligand-based Similarity Search Research Articles

Machine Learning-Based Enhanced Drug Delivery System and Its Applications – A Systematic Review

Machine Learning (ML) methods offer advanced algorithms and tools to enhance drug development. Quantitative Structure-Activity Relationship (QSAR) approaches have successfully predicted several physicochemical aspects of pharmaceuticals, including toxicity, intake, drug-drug interactions, carcinogenesis, and dispersion. ML, a branch of artificial intelligence, has demonstrated significant promise in drug development. The methods presented in this study can model non-linear databases and handle big data that is becoming more extensive and intricate. Diverse ML methodologies are currently employed for making forecasts of drug targets, modeling the framework of drug targets, forecasting binding sites, conducting ligand-based similarity searches, designing novel ligands with specific properties, creating scoring algorithms for molecular docking, constructing QSAR models for predicting biological reactions, and forecasting the pharmacokinetic and pharmacodynamic characteristics of ligands. The findings of this study illustrate the widespread utilization of ML techniques in drug discovery, suggesting a favorable outlook for these advances. These findings have the potential to facilitate further exploration of ML in connection with drug discovery and growth by researchers, learners, and pharmaceutical companies.

Repurposing Drugs for Inhibition against ALDH2 via a 2D/3D Ligand-Based Similarity Search and Molecular Simulation.

Aldehyde dehydrogenase-2 (ALDH2) is a crucial enzyme participating in intracellular aldehyde metabolism and is acknowledged as a potential therapeutic target for the treatment of alcohol use disorder and other addictive behaviors. Using previously reported ALDH2 inhibitors of Daidzin, CVT-10216, and CHEMBL114083 as reference molecules, here we perform a ligand-based virtual screening of world-approved drugs via 2D/3D similarity search methods, followed by the assessments of molecular docking, toxicity prediction, molecular simulation, and the molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) analysis. The 2D molecular fingerprinting of ECFP4 and FCFP4 and 3D molecule-shape-based USRCAT methods show good performances in selecting compounds with a strong binding behavior with ALDH2. Three compounds of Zeaxanthin (q = 0), Troglitazone (q = 0), and Sequinavir (q = +1 e) are singled out as potential inhibitors; Zeaxanthin can only be hit via USRCAT. These drugs displayed a stronger binding strength compared to the reported potent inhibitor CVT-10216. Sarizotan (q = +1 e) and Netarsudil (q = 0/+1 e) displayed a strong binding strength with ALDH2 as well, whereas they displayed a shallow penetration into the substrate-binding tunnel of ALDH2 and could not fully occupy it. This likely left a space for substrate binding, and thus they were not ideal inhibitors. The MM-PBSA results indicate that the selected negatively charged compounds from the similarity search and Vina scoring are thermodynamically unfavorable, mainly due to electrostatic repulsion with the receptor (q = -6 e for ALDH2). The electrostatic attraction with positively charged compounds, however, yielded very strong binding results with ALDH2. These findings reveal a deficiency in the modeling of electrostatic interactions (in particular, between charged moieties) in the virtual screening via the 2D/3D similarity search and molecular docking with the Vina scoring system.

Identification of a Putative α-synuclein Radioligand Using an in silico Similarity Search.

Previous studies from our lab utilized an ultra-high throughput screening method to identify compound 1 as a small molecule that binds to alpha-synuclein (α-synuclein) fibrils. The goal of the current study was to conduct a similarity search of 1 to identify structural analogs having improved in vitro binding properties for this target that could be labeled with radionuclides for both in vitro and in vivo studies for measuring α-synuclein aggregates. Using 1 as a lead compound in a similarity search, isoxazole derivative 15 was identified to bind to α-synuclein fibrils with high affinity in competition binding assays. A photocrosslinkable version was used to confirm binding site preference. Derivative 21, the iodo-analog of 15, was synthesized, and subsequently radiolabeled isotopologs [125I]21 and [11C]21 were successfully synthesized for use in in vitro and in vivo studies, respectively. [125I]21 was used in radioligand binding studies in post-mortem Parkinson's disease (PD) and Alzheimer's disease (AD) brain homogenates. In vivo imaging of an α-synuclein mouse model and non-human primates was performed with [11C]21. In silico molecular docking and molecular dynamic simulation studies for a panel of compounds identified through a similarity search, were shown to correlate with Ki values obtained from in vitro binding studies. Improved affinity of isoxazole derivative 15 for α-synuclein binding site 9 was indicated by photocrosslinking studies with CLX10. Design and successful (radio)synthesis of iodo-analog 21 of isoxazole derivative 15 enabled further in vitro and in vivo evaluation. Kd values obtained in vitro with [125I]21 for α-synuclein and Aβ42 fibrils were 0.48 ± 0.08 nM and 2.47 ± 1.30 nM, respectively. [125I]21 showed higher binding in human postmortem PD brain tissue compared with AD tissue, and low binding in control brain tissue. Lastly, in vivo preclinical PET imaging showed elevated retention of [11C]21 in PFF-injected mouse brain. However, in PBS-injected control mouse brain, slow washout of the tracer indicates high non-specific binding. [11C]21 showed high initial brain uptake in a healthy non-human primate, followed by fast washout that may be caused by rapid metabolic rate (21% intact [11C]21 in blood at 5 min p.i.). Through a relatively simple ligand-based similarity search, we identified a new radioligand that binds with high affinity (<10 nM) to α-synuclein fibrils and PD tissue. Although the radioligand has suboptimal selectivity for α-synuclein towards Aβ and high non-specific binding, we show here that a simple in silico approach is a promising strategy to identify novel ligands for target proteins in the CNS with the potential to be radiolabeled for PET neuroimaging studies.

Screening of β1- and β2-Adrenergic Receptor Modulators through Advanced Pharmacoinformatics and Machine Learning Approaches.

Cardiovascular diseases (CDs) are a major concern in the human race and one of the leading causes of death worldwide. β-Adrenergic receptors (β1-AR and β2-AR) play a crucial role in the overall regulation of cardiac function. In the present study, structure-based virtual screening, machine learning (ML), and a ligand-based similarity search were conducted for the PubChem database against both β1- and β2-AR. Initially, all docked molecules were screened using the threshold binding energy value. Molecules with a better binding affinity were further used for segregation as active and inactive through ML. The pharmacokinetic assessment was carried out on molecules retained in the above step. Further, similarity searching of the ChEMBL and DrugBank databases was performed. From detailed analysis of the above data, four compounds for each of β1- and β2-AR were found to be promising in nature. A number of critical ligand-binding amino acids formed potential hydrogen bonds and hydrophobic interactions. Finally, a molecular dynamics (MD) simulation study of each molecule bound with the respective target was performed. A number of parameters obtained from the MD simulation trajectories were calculated and substantiated the stability between the protein-ligand complex. Hence, it can be postulated that the final molecules might be crucial for CDs subjected to experimental validation.

Potent antiplasmodial alkaloids from the rhizobacterium Pantoea agglomerans as hemozoin modulators

Global health concern regarding malaria has increased since the first report of artemisinin-resistant Plasmodium falciparum (Pf) two decades ago. The current therapies suffer various drawbacks such as low efficacy and significant side effects, alarming for an urgent need of more effective and less toxic drugs with higher patient compliance. Chemical entities with natural origins become progressively attractive as new drug leads due to their structural diversity and bio-compatibility. This study initially aimed at the targeted isolation of hydroxyquinoline derivatives following our published genomics and metabolomics study of Pantoea agglomerans (Pa). Fermentation of Pa on a pre-selected medium followed by chromatographic isolation, NMR and HRMS analyses led to the characterisation of one new hydroxyquinoline alkaloid together with another six known congeners and two known hydroxyquinolone derivatives. When screened for their antimalarial activity by high throughput screening against asexual blood-stage parasites, almost all compounds showed potent and selective sub-micromolar activities. Computational investigation was performed to identify the antiplasmodial potential targets. Ligand-based similarity search predicted the tested compounds to act as hemozoin inhibitors. Computational target identification results were further validated by competitive hemozoin inhibitory properties of hydroxyquinoline and hydroxyquinolone derivatives in vitro. The overall results suggest this natural scaffold is of potential to be developed as antimalarial drug lead.

Molecular screening of glycyrrhizin-based inhibitors against ACE2 host receptor of SARS-CoV-2.

The interaction between SARS-CoV-2 Spike protein and angiotensin-converting enzyme 2 (ACE2) is essential to viral attachment and the subsequent fusion process. Interfering with this event represents an attractive avenue for the development of therapeutics and vaccine development. Here, a hybrid approach of ligand- and structure-based virtual screening techniques were employed to disclose similar analogues of a reported antiviral phytochemical, glycyrrhizin, targeting the blockade of ACE2 interaction with the SARS-CoV-2 Spike. A ligand-based similarity search using a stringent cut-off revealed 40 FDA-approved compounds in DrugBank. These filtered hits were screened against ACE2 using a blind docking approach to determine the natural binding tendency of the compounds with ACE2. Three compounds, deslanoside, digitoxin, and digoxin, were reported to show strong binding with ACE2. These compounds bind at the H1-H2 binding pocket, in a manner similar to that of glycyrrhizin which was used as a control. To achieve consistency in the docking results, docking calculations were performed via two sets of docking software that predicted binding energy as ACE2-Deslanoside (AutoDock, −10.3 kcal/mol and DockThor, −9.53 kcal/mol), ACE2-Digitoxin (AutoDock, −10.6 kcal/mol and DockThor, −8.84 kcal/mol), and ACE2-Digoxin (AutoDock, −10.6 kcal/mol and DockThor, −8.81 kcal/mol). The docking results were validated by running molecular simulations in aqueous solution that demonstrated the stability of ACE2 with no major conformational changes in the ligand original binding mode (~ 2 Å average RMSD). Binding interactions remained quite stable with an increased potential for getting stronger as the simulation proceeded. MMGB/PBSA binding free energies were also estimated and these supported the high stability of the complexes compared to the control (~ −50 kcal/mol net MMGB/PBSA binding energy versus ~ −30 kcal/mol). Collectively, the data demonstrated that the compounds shortlisted in this study might be subjected to experimental evaluation to uncover their real blockade capacity of SARS-CoV-2 host ACE2 receptor.Graphical abstract

Rational design of potent anti-COVID-19 main protease drugs: An extensive multi-spectrum in silico approach

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) as a novel coronavirus and the etiological agent of global pandemic coronavirus disease (COVID-19) requires quick development of potential therapeutic strategies. Computer aided drug design approaches are highly efficient in identifying promising drug candidates among an available pool of biological active antivirals with safe pharmacokinetics. The main protease (MPro) enzyme of SARS-CoV-2 is considered key in virus production and its crystal structures are available at excellent resolution. This marks the enzyme as a good starting receptor to conduct an extensive structure-based virtual screening (SBVS) of ASINEX antiviral library for the purpose of uncovering valuable hits against SARS-CoV-2 MPro. A compound hit (BBB_26580140) was stand out in the screening process, as opposed to the control, as a potential inhibitor of SARS-CoV-2 MPro based on a combined approach of SBVS, drug likeness and lead likeness annotations, pharmacokinetics, molecular dynamics (MD) simulations, and end point MM-PBSA binding free energy methods. The lead was further used in ligand-based similarity search (LBSS) that found 33 similar compounds from the ChEMBL database. A set of three compounds (SCHEMBL12616233, SCHEMBL18616095, and SCHEMBL20148701), based on their binding affinity for MPro, was selected and analyzed using extensive MD simulation, hydrogen bond profiling, MM-PBSA, and WaterSwap binding free energy techniques. The compounds conformation with MPro show good stability after initial within active cavity moves, a rich intermolecular network of chemical interactions, and reliable relative and absolute binding free energies. Findings of the study suggested the use of BBB_26580140 lead and its similar analogs to be explored in vivo which might pave the path for rational drug discovery against SARS-CoV-2 MPro.

The ubiquinone synthesis pathway is a promising drug target for Chagas disease.

Chagas disease is caused by infection with the protozoan parasite Trypanosoma cruzi (T. cruzi). It was originally a Latin American endemic health problem, but now is expanding worldwide as a result of increasing migration. The currently available drugs for Chagas disease, benznidazole and nifurtimox, provoke severe adverse effects, and thus the development of new drugs is urgently required. Ubiquinone (UQ) is essential for respiratory chain and redox balance in trypanosomatid protozoans, therefore we aimed to provide evidence that inhibitors of the UQ biosynthesis have trypanocidal activities. In this study, inhibitors of the human COQ7, a key enzyme of the UQ synthesis, were tested for their trypanocidal activities because they were expected to cross-react and inhibit trypanosomal COQ7 due to their genetic homology. We show the trypanocidal activity of a newly found human COQ7 inhibitor, an oxazinoquinoline derivative. The structurally similar compounds were selected from the commercially available compounds by 2D and 3D ligand-based similarity searches. Among 38 compounds selected, 12 compounds with the oxazinoquinoline structure inhibited significantly the growth of epimastigotes of T. cruzi. The most effective 3 compounds also showed the significant antitrypanosomal activity against the mammalian stage of T. cruzi at lower concentrations than benznidazole, a commonly used drug today. We found that epimastigotes treated with the inhibitor contained reduced levels of UQ9. Further, the growth of epimastigotes treated with the inhibitors was partially rescued by UQ10 supplementation to the culture medium. These results suggest that the antitrypanosomal mechanism of the oxazinoquinoline derivatives results from inhibition of the trypanosomal UQ synthesis leading to a shortage of the UQ pool. Our data indicate that the UQ synthesis pathway of T. cruzi is a promising drug target for Chagas disease.

Ligand-based Discovery of Novel Small Molecule Inhibitors of RON Receptor Tyrosine Kinase.

RON (Recepteur d'Origine Nantais) receptor tyrosine kinase is a promising target for anti-cancer therapeutics. The aim of this study was to identify new RON inhibitors using virtual screening methods. To this end, a ligand-based virtual screening approach was employed for screening of ZINC database on the homology model of RON receptor. All the selected hits were inspected in terms of drug-likeness, ADME properties, and toxicity profiles. Ligand-based similarity searches along with further filtering criteria led to the identification of two compounds, TKI1 and TKI2 that were evaluated using in vitro cell-based RON inhibition assays. The results showed that TKI1 and TKI2 could reduce phosphorylation of RON. Both compounds showed inhibitory activity of the downstream mTOR pathway with no apparent effects on other signaling mediators in a dose-dependent manner. These compounds can provide a basis for developing novel anti-RON inhibitors applicable to cancer therapy using medicinal chemistry-oriented optimization strategies.

Synthesis, biological evaluation and QSAR studies of new thieno[2,3-d]pyrimidin-4(3H)-one derivatives as antimicrobial and antifungal agents

A series of new thieno[2,3-d]pyrimidin-4(3H)-one derivatives were synthesized and evaluated for their activity against four gram-positive and four gram-negative bacterial and eight fungal species. The majority of the compounds exhibited excellent antimicrobial and antifungal activity, being more potent than the control compounds. Compound 22, bearing a m-methoxyphenyl group and an ethylenediamine side chain anchored at C-2 of the thienopyrimidinone core, is the most potent antibacterial compound with broad antimicrobial activity with MIC values in the range of 0.05–0.13 mM, being 6 to 15 fold more potent than the controls, streptomycin and ampicillin. Furthermore, compounds 14 and 15 which bear a p-chlorophenyl and m-methoxyphenyl group, respectively, and share a 2-(2-mercaptoethoxy)ethan-1-ol side chain showed the best antifungal activity, being 10–15 times more potent than ketoconazole or bifonazole with MIC values 0.013–0.026 and 0.027 mM, respectively. Especially in the case of compound 15 the low MIC values were accompanied by excellent MFC values ranging from 0.056 to 0.058 mM. Evaluation of toxicity in vitro on HFL-1 human embryonic primary cells and in vivo in the nematode C. elegans revealed no toxic effects for both compounds 15 and 22 tested at the MIC concentrations. Ligand-based similarity search and molecular docking predicted that the antibacterial activity of analogue 22 is related to inhibition of the topoisomerase II DNA gyrase enzyme and the antifungal activity of compound 15 to CYP51 lanosterol demethylase enzyme. R-Group analysis as a means of computational structure activity relationship tool, highlighted the compounds’ crucial pharmacophore features and their impact on the antibacterial and antifungal activity. The presence of a N-methyl piperidine ring fused to the thienopyrimidinone core plays an important role in both activities.

Virtual Screening in the Cloud: How Big Is Big Enough?

Virtual screening is a standard tool in Computer-Assisted Drug Design (CADD). Early in a project, it is typical to use ligand-based similarity search methods to find suitable hit molecules. However, the number of compounds which can be screened and the time required are usually limited by computational resources. We describe here a high-throughput virtual screening project using 3D similarity (FastROCS) and automated evaluation workflows on Orion, a cloud computing platform. Cloud resources make this approach fully scalable and flexible, allowing the generation and search of billions of virtual molecules, and give access to an explicit 3D virtual chemistry space not available before. We discuss the impact of the size of the search space with respect to finding novel chemical hits and the size of the required hit list, as well as computational and economical aspects of resource scaling.

Putative histidine kinase inhibitors with antibacterial effect against multi-drug resistant clinical isolates identified by in vitro and in silico screens.

Novel antibacterials are urgently needed to address the growing problem of bacterial resistance to conventional antibiotics. Two-component systems (TCS) are widely used by bacteria to regulate gene expression in response to various environmental stimuli and physiological stress and have been previously proposed as promising antibacterial targets. TCS consist of a sensor histidine kinase (HK) and an effector response regulator. The HK component contains a highly conserved ATP-binding site that is considered to be a promising target for broad-spectrum antibacterial drugs. Here, we describe the identification of putative HK autophosphorylation inhibitors following two independent experimental approaches: in vitro fragment-based screen via differential scanning fluorimetry and in silico structure-based screening, each followed up by the exploration of analogue compounds as identified by ligand-based similarity searches. Nine of the tested compounds showed antibacterial effect against multi-drug resistant clinical isolates of bacterial pathogens and include three novel scaffolds, which have not been explored so far in other antibacterial compounds. Overall, putative HK autophosphorylation inhibitors were found that together provide a promising starting point for further optimization as antibacterials.

Case-specific performance of MM-PBSA, MM-GBSA, and SIE in virtual screening

In drug discovery the reliable prediction of binding free energies is of crucial importance. Methods that combine molecular mechanics force fields with continuum solvent models have become popular because of their high accuracy and relatively good computational efficiency. In this research we studied the performance of molecular mechanics generalized Born surface area (MM-GBSA), molecular mechanics Poisson–Boltzmann surface area (MM-PBSA), and solvated interaction energy (SIE) both in their virtual screening efficiency and their ability to predict experimentally determined binding affinities for five different protein targets. The protein-ligand complexes were derived with two different approaches important in virtual screening: molecular docking and ligand-based similarity search methods. The results show significant differences between the different binding energy calculation methods. However, the length of the molecular dynamics simulation was not of crucial importance for accuracy of results.

HybridDock: A Hybrid Protein-Ligand Docking Protocol Integrating Protein- and Ligand-Based Approaches.

Structure-based molecular docking and ligand-based similarity search are two commonly used computational methods in computer-aided drug design. Structure-based docking tries to utilize the structural information on a drug target like protein, and ligand-based screening takes advantage of the information on known ligands for a target. Given their different advantages, it would be desirable to use both protein- and ligand-based approaches in drug discovery when information for both the protein and known ligands is available. Here, we have presented a general hybrid docking protocol, referred to as HybridDock, to utilize both the protein structures and known ligands by combining the molecular docking program MDock and the ligand-based similarity search method SHAFTS, and evaluated our hybrid docking protocol on the CSAR 2013 and 2014 exercises. The results showed that overall our hybrid docking protocol significantly improved the performance in both binding affinity and binding mode predictions, compared to the sole MDock program. The efficacy of the hybrid docking protocol was further confirmed using the combination of DOCK and SHAFTS, suggesting an alternative docking approach for modern drug design/discovery.

Prediction of protein targets of kinetin using in silico and in vitro methods: a case study on spinach seed germination mechanism.

Kinetin, a cytokinin which promotes seed germination by inhibiting the action of abscisic acid, is an important molecule known to trigger various molecular mechanisms by interacting with an array of proteins shown from experimental observations in various model organisms. We report here the prediction of most probable protein targets of kinetin from spinach proteome using in silico approaches. Inverse docking and ligand-based similarity search was performed using kinetin as molecule. The former method prioritized six spinach proteins, whereas the latter method provided a list of protein targets retrieved from several model organisms. The most probable protein targets were selected by comparing the rank list of docking and ligand similarity methods. Both of these methods prioritized chitinase as the most probable protein target (ΔG pred = 5.064kcal/mol) supported by the experimental structure of yeast chitinase 1 complex with kinetin (PDB: 2UY5) and Gliocladium roseum chitinase complex with 3,7-dihydro-1,3,7-trimethyl-1H-purine-2,6-dione (caffeine; 3G6M) which bears a 3D similarity of 0.43 with kinetin. An in vitro study to evaluate the effect of kinetin on spinach seed germination indicated that a very low concentration of kinetin (0.5mg/l) did not show a significant effect compared to control in inducing seed germination process. Further, higher levels of kinetin (>0.5mg/l) constituted an antagonist effect on spinach seed germination. It is anticipated that kinetin may have a molecular interaction with prioritized protein targets synthesized during the seed germination process and reduces growth. Thus, it appears that kinetin may not be a suitable hormone for enhancing spinach seed germination in vitro.

Discovery of novel small-molecule compounds with selective cytotoxicity for Burkitt's lymphoma cells using 3D ligand-based virtual screening.

We describe a ligand-based approach towards compounds with more specific targeting for Burkitt’s lymphoma. Using three-dimensional ligand-based similarity searches and a previously described hit compound, we have identified six compounds that are chemically different but with similar spatial conformations. Biological evaluation revealed that one compound has better growth inhibition and improved selectivity towards Burkitt’s lymphoma cells than the query compound. However, initial mechanism-of-action studies show a different target profile in comparison with the previous hit compound, which does not involve the inhibition of the proteasome or the NFκB pathway. The data from this study provide a solid basis for further efforts in the search for selective agents against Burkitt’s lymphoma.

Combination of 2D/3D ligand-based similarity search in rapid virtual screening from multimillion compound repositories. Selection and biological evaluation of potential PDE4 and PDE5 inhibitors.

Rapid in silico selection of target focused libraries from commercial repositories is an attractive and cost effective approach. If structures of active compounds are available rapid 2D similarity search can be performed on multimillion compound databases but the generated library requires further focusing by various 2D/3D chemoinformatics tools. We report here a combination of the 2D approach with a ligand-based 3D method (Screen3D) which applies flexible matching to align reference and target compounds in a dynamic manner and thus to assess their structural and conformational similarity. In the first case study we compared the 2D and 3D similarity scores on an existing dataset derived from the biological evaluation of a PDE5 focused library. Based on the obtained similarity metrices a fusion score was proposed. The fusion score was applied to refine the 2D similarity search in a second case study where we aimed at selecting and evaluating a PDE4B focused library. The application of this fused 2D/3D similarity measure led to an increase of the hit rate from 8.5% (1st round, 47% inhibition at 10 µM) to 28.5% (2nd round at 50% inhibition at 10 µM) and the best two hits had 53 nM inhibitory activities.

Discovery and structure–activity analysis of selective estrogen receptor modulators via similarity-based virtual screening

A number of selective estrogen receptor modulators (SERMs) were discovered from the SPECS database via a simple protocol. Based on two reference SERMs we identified via structure-based virtual screening previously, ligand-based similarity search and molecular docking filtering were conducted to identify novel SERMs from SPECS library. Among the 36 purchased compounds, 21 were confirmed to be active by in vitro assays, and 10 showed dual profile properties, namely as antagonists of ERα and agonists of ERβ. The anti-proliferative potency of these ligands was also evaluated against MCF-7 cell lines. Further investigation of the anti-proliferative mechanism of compound 3a suggested that it induced a G1 cell cycle arrest in ERα positive MCF-7 cell through ERα mediated cyclin D1 down-regulation.

PLS-DA - Docking Optimized Combined Energetic Terms (PLSDA-DOCET) Protocol: A Brief Evaluation

Docking studies have become popular approaches in drug design, where the binding energy of the ligand in the active site of the protein is estimated by a scoring function. Many promising techniques were developed to enhance the performance of scoring functions including the fusion of multiple scoring functions outcomes into a so-called consensus scoring function. Hereby, we evaluated the target oriented consensus technique using the energetic terms of several scoring functions. The approach was denoted PLSDA-DOCET. Optimization strategies for consensus energetic terms and scoring functions based on ROC metric were compared to classical rigid docking and to ligand-based similarity search methods comprising 2D fingerprints and ROCS. The ROCS results indicate large performance variations depending on the biological target. The AUC-based strategy of PLSDA-DOCET outperformed the other docking approaches regarding simple retrieval and scaffold-hopping. The superior performance of PLSDA-DOCET protocol relative to single and combined scoring functions was validated on an external test set. We found a relative low mean correlation of the ranks of the chemotypes retrieved by the PLSDA-DOCET protocol and all the other methods employed here.

Electron Density Fingerprints (EDprints): Virtual Screening Using Assembled Information of Electron Density

We have designed a method to encode properties related to the electron densities of molecules (calculated (1)H and (13)C NMR shifts and atomic partial charges) in molecular fingerprints (EDprints). EDprints was evaluated in terms of their retrospective virtual screening accuracy against the Directory of Useful Decoys (DUD) and compared to the established ligand-based similarity search methods MOLPRINT 2D and FCFP-4. Although there are no significant differences in the overall virtual screening accuracies of the three methods, specific examples highlight interesting differences between the new EDprints fingerprint method and the atom-centered circular fingerprint methods of MOLPRINT 2D and FCFP-4. On one hand, EDprints similarity searches can be biased by the molecular protonation state, especially when reference ligands contain multiple ionizable groups. On the other hand, EDprints models are more robust toward subtle rearrangements of chemical groups and more suitable for screening against reference molecules with fused ring systems than MOLPRINT 2D and FCFP-4. EDprints is furthermore the fastest method under investigation in comparing fingerprints (average 56-233-fold increase in speed), which makes it highly suitable for all-against-all similarity searches and for repetitive virtual screening against large chemical databases of millions of compounds.