Large-scale evaluation of log P predictors: local corrections may compensate insufficient accuracy and need of experimentally testing every other compound

Discovery of a Novel Orally Active, Selective LPA Receptor Type 1 Antagonist, 4-(4-(2-Isopropylphenyl)-4-((2-methoxy-4-methylphenyl)carbamoyl)piperidin-1-yl)-4-oxobutanoic Acid, with a Distinct Molecular Scaffold

Lysophosphatidic acid receptor 1 (LPAR1) antagonists show promise as potentially novel antifibrotic treatments. In a human LPAR1 β-arrestin recruitment-based high-throughput screening campaign, we identified urea 19 as a hit with a LPAR1 IC50 value of 5.0 μM. Hit-to-lead activities revealed that one of the urea nitrogen atoms can be replaced by carbon and establish the corresponding phenylacetic amide as a lead structure for further optimization. Medicinal chemistry efforts led to the discovery of piperidine 18 as a potent and selective LPAR1 antagonist with oral activity in a mouse model of LPA-induced skin vascular leakage. The molecular scaffold of 18 shares no obvious structural similarity with any other LPAR1 antagonist disclosed so far.

Evaluation of Free Online ADMET Tools for Academic or Small Biotech Environments

For a new molecular entity (NME) to become a drug, it is not only essential to have the right biological activity also be safe and efficient, but it is also required to have a favorable pharmacokinetic profile including toxicity (ADMET). Consequently, there is a need to predict, during the early stages of development, the ADMET properties to increase the success rate of compounds reaching the lead optimization process. Since Lipinski's rule of five, the prediction of pharmacokinetic parameters has evolved towards the current in silico tools based on empirical approaches or molecular modeling. The commercial specialized software for performing such predictions, which is usually costly, is, in many cases, not among the possibilities for research laboratories in academia or at small biotech companies. Nevertheless, in recent years, many free online tools have become available, allowing, more or less accurately, for the prediction of the most relevant pharmacokinetic parameters. This paper studies 18 free web servers capable of predicting ADMET properties and analyzed their advantages and disadvantages, their model-based calculations, and their degree of accuracy by considering the experimental data reported for a set of 24 FDA-approved tyrosine kinase inhibitors (TKIs) as a model of a research project.

A High-Throughput LC-MS/MS Method for the Simultaneous Quantification of Twenty-Seven Drug Molecules in Ocular Tissues

The purpose of this study was to develop a validated LC-MS/MS analytical method for the simultaneous analysis of a large cassette containing a wide range of drug substances with positive, negative, or neutral charge and further apply the method to assess octanol partition coefficient and eye tissue recovery of the drug cassette. A twenty-seven-drug cassette (N-in-one) including beta blockers, NSAIDs, and corticosteroids that range from extremely hydrophilic (sotalol) to very hydrophobic (triamcinolone hexacetanide) was used to develop an LC-MS/MS assay using QTrap 4500. An LC-MS/MS method based on gradient elution, with an eighteen-minute run time including equilibration time, was developed and validated for the rapid and simultaneous quantitation of drugs with a wide range of lipophilicities. Scheduled multiple reaction monitoring was used to maximize the scan time for each peak, ensuring sufficient scans. Method validation included lower limit of quantitation (LLOQ) and intra- and inter-day reproducibility. The LLOQ ranged from 0.5 (sotalol) to 40 fmols (dexamethasone) on column with a %RSD < 20%. The method was tested by measuring octanol:water and octanol:buffer (PBS, pH 7.4) partition coefficients and by quantitation of the drug cassette extracted from rabbit aqueous humor and cornea. Measured partition coefficients correlated positively with predicted values (r²=0.5-0.7). Drug recovery was ≥ 79% from aqueous humor and between 61 and 67% on average from cornea. A rapid, sensitive LC-MS/MS method suitable for N-in-one drug delivery screening was developed for simultaneous quantification of twenty-seven drugs in aqueous solutions and eye tissues.

Highly Hydrophilic and Lipophilic Derivatives of Bile Salts

Lipophilicity of 15 derivatives of sodium cholate, defined by the octan-1-ol/water partition coefficient (log P), has been theoretically determined by the Virtual log P method. These derivatives bear highly hydrophobic or highly hydrophilic substituents at the C3 position of the steroid nucleus, being linked to it through an amide bond. The difference between the maximum value of log P and the minimum one is enlarged to 3.5. The partition coefficient and the critical micelle concentration (cmc) are tightly related by a double-logarithm relationship (VirtuallogP=−(1.00±0.09)log(cmcmM)+(2.79±0.09)), meaning that the Gibbs free energies for the transfer of a bile anion from water to either a micelle or to octan-1-ol differ by a constant. The equation also means that cmc can be used as a measurement of lipophilicity. The demicellization of the aggregates formed by three derivatives of sodium cholate bearing bulky hydrophobic substituents has been studied by surface tension and isothermal titration calorimetry. Aggregation numbers, enthalpies, free energies, entropies, and heat capacities, ΔC_P,demic, were obtained. ΔC_P,demic, being positive, means that the interior of the aggregates is hydrophobic.

Nucleolipids of the Nucleoside Antibiotics Formycins A and B: Synthesis and Biomedical Characterization Particularly Using Glioblastoma Cells

Two lipophilic derivatives of formycin A (1) and formycin B (5) carrying an O-2',3'-(ethyl levulinate) ketal group have been prepared. These were base-alkylated at N(1) (for 1) and N(1) and N(6) (for 5) with both isopentenyl and all-trans-farnesyl residues. Upon the prenylation, side reactions were observed, resulting in the formation of nucleolipids with a novel tricyclic nucleobase (→4a, 4b). In the case of formycin B, O-2',3'-(ethyl levulinate) (6) farnesylation gave the double prenylated nucleolipid 7. All new compounds were characterized by ¹ H-, ¹³ C-, UV/VIS and fluorescence spectroscopy, by ESI-MS spectrometry and/or by elemental analysis. Log P determinations between water and octanol as well as water and cyclohexane of a selection of compounds allowed qualitative conclusions concerning their potential blood-brain barrier passage efficiency. All compounds were investigated in vitro with respect to their cytotoxic activity toward rat malignant neuroectodermal BT4Ca as well as against a series of human glioblastoma cell lines (GOS 3, U-87 MG and GBM 2014/42). In order to differentiate between anticancer and side effects of the novel nucleolipids, we also studied their activity on PMA-differentiated human THP-1 macrophages. Here, we show that particularly the formycin A derivative 3b possesses promising antitumor properties in several cancer cell lines with profound cytotoxic effects partly on human glioblastoma cells, with a higher efficacy than the chemotherapeutic drug 5-fluorouridine.

Guanosine Nucleolipids: Synthesis, Characterization, Aggregation and X-Ray Crystallographic Identification of Electricity-Conducting G-Ribbons

The lipophilization of β-d-riboguanosine (1) with various symmetric as well as asymmetric ketones is described (→3a-3f). The formation of the corresponding O-2',3'-ketals is accompanied by the appearance of various fluorescent by-products which were isolated chromatographically as mixtures and tentatively analyzed by ESI-MS spectrometry. The mainly formed guanosine nucleolipids were isolated and characterized by elemental analyses, ¹ H-, ¹³ C-NMR and UV spectroscopy. For a drug profiling, static topological polar surface areas as well as ¹⁰ logP_OW values were calculated by an increment-based method as well as experimentally for the systems 1-octanol-H₂ O and cyclohexane-H₂ O. The guanosine-O-2',3'-ketal derivatives 3b and 3a could be crystallized in (D₆ )DMSO - the latter after one year of standing at ambient temperature. X-ray analysis revealed the formation of self-assembled ribbons consisting of two structurally similar 3b nucleolipid conformers as well as integrated (D₆ )DMSO molecules. In the case of 3a ⋅ DMSO, the ribbon is formed by a single type of guanosine nucleolipid molecules. The crystalline material 3b ⋅ DMSO was further analyzed by differential scanning calorimetry (DSC) and temperature-dependent polarization microscopy. Crystallization was also performed on interdigitated electrodes (Au, distance, 5 μm) and visualized by scanning electron microscopy. Resistance and amperage measurements clearly demonstrate that the electrode-bridging 3b crystals are electrically conducting. All O-2',3'-guanosine ketals were tested on their cytostatic/cytotoxic activity towards phorbol 12-myristate 13-acetate (PMA)-differentiated human THP-1 macrophages as well as against human astrocytoma/oligodendroglioma GOS-3 cells and against rat malignant neuroectodermal BT4Ca cells.

Synthesis of New Potential Lipophilic Co-Drugs of 2-Chloro-2'-deoxyadenosine (Cladribine, 2-CdA, Mavenclad®, Leustatin®) and 6-Azauridine (z6 U) with Valproic Acid

2-Chloro-2'-deoxyadenosine (cladribine, 1) was acylated with valproic acid (2) under various reaction conditions yielding 2-chloro-2'-deoxy-3',5'-O-divalproyladenosine (3) as well as the 3'-O- and 5'-O-monovalproylated derivatives, 2-chloro-2'-deoxy-3'-O-valproyladenosine (4) and 2-chloro-2'-deoxy-5'-O-valproyladenosine (5), as new co-drugs. In addition, 6-azauridine-2',3'-O-(ethyl levulinate) (8) was valproylated at the 5'-OH group (→9). All products were characterized by ¹ H- and ¹³ C-NMR spectroscopy and ESI mass spectrometry. The structure of the by-product 6 (N-cyclohexyl-N-(cyclohexylcarbamoyl)-2-propylpentanamide), formed upon valproylation of cladribine in the presence of N,N-dimethylaminopyridine and dicyclohexylcarbodiimide, was analyzed by X-ray crystallography. Cladribine as well as its valproylated co-drugs were tested upon their cancerostatic/cancerotoxic activity in human astrocytoma/oligodendroglioma GOS-3 cells, in rat malignant neuro ectodermal BT4Ca cells, as well as in phorbol-12-myristate 13-acetate (PMA)-differentiated human THP-1 macrophages. The most important result of these experiments is the finding that only the 3'-O-valproylated derivative 4 exhibits a significant antitumor activity while the 5'-O- as well as the 3',5'-O-divalproylated cladribine derivatives 3 and 5 proved to be inactive.

Dissecting Machine-Learning Prediction of Molecular Activity: Is an Applicability Domain Needed for Quantitative Structure-Activity Relationship Models Based on Deep Neural Networks?

Deep neural networks (DNNs) are the major drivers of recent progress in artificial intelligence. They have emerged as the machine-learning method of choice in solving image and speech recognition problems, and their potential has raised the expectation of similar breakthroughs in other fields of study. In this work, we compared three machine-learning methods-DNN, random forest (a popular conventional method), and variable nearest neighbor (arguably the simplest method)-in their ability to predict the molecular activities of 21 in vivo and in vitro data sets. Surprisingly, the overall performance of the three methods was similar. For molecules with structurally close near neighbors in the training sets, all methods gave reliable predictions, whereas for molecules increasingly dissimilar to the training molecules, all three methods gave progressively poorer predictions. For molecules sharing little to no structural similarity with the training molecules, all three methods gave a nearly constant value-approximately the average activity of all training molecules-as their predictions. The results confirm conclusions deduced from analyzing molecular applicability domains for accurate predictions, i.e., the most important determinant of the accuracy of predicting a molecule is its similarity to the training samples. This highlights the fact that even in the age of deep learning, developing a truly high-quality model relies less on the choice of machine-learning approach and more on the availability of experimental efforts to generate sufficient training data of structurally diverse compounds. The results also indicate that the distance to training molecules offers a natural and intuitive basis for defining applicability domains to flag reliable and unreliable quantitative structure-activity relationship predictions.

Combinatorial Synthesis of New Pyrimidine- and Purine-β-d-Ribonucleoside Nucleolipids: Their Distribution Between Aqueous and Organic Phases and Their In Vitro Activity Against Human- and Rat Glioblastoma Cells In Vitro

Two series of nucleolipids, O-2',3'-heptanylidene- as well as O-2',3'-undecanylidene ketals of six β-d-ribonucleosides (type A) and partly N-farnesyl derivatives thereof (type B) were prepared in a combinatorial manner. All novel compounds were characterized by elemental analysis and/or ESI mass spectrometry and by UV-, ¹ H-, and ¹³ C-NMR spectroscopy. Conformational parameters of the nucleosides and nucleolipids were calculated from various ³ J(H,H), ³ J(¹ H,¹³ C), and ⁵ J(F,H) coupling constants. For a drug profiling, the parent nucleosides and their lipophilic derivatives were studied with respect to their distribution (log P) between water and n-octanol as well as water and cyclohexane. From these data, qualitative conclusions were drawn concerning their possible blood-brain barrier passage efficiency. Moreover, nucleolipids were characterized by their molecular descriptor amphiphilic ratio (a.r.), which describes the balance between the hydrophilicity of the nucleoside headgroup and the lipophilicity of the lipid tail. All compounds were investigated in vitro with respect to their cytostatic/cytotoxic activity toward human glioblastoma (GOS 3) as well as rat malignant neuroectodermal BT4Ca cell lines in vitro. In order to differentiate between anticancer and side-effects of the novel nucleolipids, they were also studied on their activity on differentiated human THP-1 macrophages.

Ameliorated or Acquired Cytostatic/Cytotoxic Properties of Nucleosides by Lipophilization

A series of nucleolipids, containing one of the β-D-ribonucleosides 5-fluorouridine, 6-azauridine, uridine, or 5-methyluridine were lipophilized, either at the O-2',3'-position and/or at N(3) of the nucleobase with a large variety of hydrophobic residues. The resulting nucleolipids as well as the parent nucleosides and the lipid precursors were investigated in vitro with respect to their antitumor activity towards i) ten human tumor cell lines from the NCI 60 panel and ii) partly against three further tumor cell lines, namely a) human astrocytoma/oligodendro glioma GOs-3, b) rat malignantneuroectodermal BT4Ca, and c) differentiated human THP-1 macrophages. Inspection of the doseresponse curves allows two main conclusions concerning lipid determinants lending the corresponding nucleoside an ameliorated or an acquired antitumor activity: i) introduction of either a symmetrical O-2',3'-nonadecylidene ketal group or introduction of an O-2',3'-ethyl levulinate moiety plus an N(3)-farnesyl group leads often to nucleolipids with significant cytostatic/cytotoxic properties; ii) for the two canonical and non-toxic nucleosides uridine and 5-methyluridine, the condensation with also non-toxic lipids gives nucleolipids with a pronounced antitumor activity.

A renaissance of neural networks in drug discovery

INTRODUCTION

Neural networks are becoming a very popular method for solving machine learning and artificial intelligence problems. The variety of neural network types and their application to drug discovery requires expert knowledge to choose the most appropriate approach.

AREAS COVERED

In this review, the authors discuss traditional and newly emerging neural network approaches to drug discovery. Their focus is on backpropagation neural networks and their variants, self-organizing maps and associated methods, and a relatively new technique, deep learning. The most important technical issues are discussed including overfitting and its prevention through regularization, ensemble and multitask modeling, model interpretation, and estimation of applicability domain. Different aspects of using neural networks in drug discovery are considered: building structure-activity models with respect to various targets; predicting drug selectivity, toxicity profiles, ADMET and physicochemical properties; characteristics of drug-delivery systems and virtual screening.

EXPERT OPINION

Neural networks continue to grow in importance for drug discovery. Recent developments in deep learning suggests further improvements may be gained in the analysis of large chemical data sets. It's anticipated that neural networks will be more widely used in drug discovery in the future, and applied in non-traditional areas such as drug delivery systems, biologically compatible materials, and regenerative medicine.

Lowe D, Williams AJ. The development of models to predict melting and pyrolysis point data associated with several hundred thousand compounds mined from PATENTS

BACKGROUND

Melting point (MP) is an important property in regards to the solubility of chemical compounds. Its prediction from chemical structure remains a highly challenging task for quantitative structure-activity relationship studies. Success in this area of research critically depends on the availability of high quality MP data as well as accurate chemical structure representations in order to develop models. Currently, available datasets for MP predictions have been limited to around 50k molecules while lots more data are routinely generated following the synthesis of novel materials. Significant amounts of MP data are freely available within the patent literature and, if it were available in the appropriate form, could potentially be used to develop predictive models.

RESULTS

We have developed a pipeline for the automated extraction and annotation of chemical data from published PATENTS. Almost 300,000 data points have been collected and used to develop models to predict melting and pyrolysis (decomposition) points using tools available on the OCHEM modeling platform (http://ochem.eu). A number of technical challenges were simultaneously solved to develop models based on these data. These included the handing of sparse data matrices with >200,000,000,000 entries and parallel calculations using 32 × 6 cores per task using 13 descriptor sets totaling more than 700,000 descriptors. We showed that models developed using data collected from PATENTS had similar or better prediction accuracy compared to the highly curated data used in previous publications. The separation of data for chemicals that decomposed rather than melting, from compounds that did undergo a normal melting transition, was performed and models for both pyrolysis and MPs were developed. The accuracy of the consensus MP models for molecules from the drug-like region of chemical space was similar to their estimated experimental accuracy, 32 °C. Last but not least, important structural features related to the pyrolysis of chemicals were identified, and a model to predict whether a compound will decompose instead of melting was developed.

CONCLUSIONS

We have shown that automated tools for the analysis of chemical information have reached a mature stage allowing for the extraction and collection of high quality data to enable the development of structure-activity relationship models. The developed models and data are publicly available at http://ochem.eu/article/99826.

Nucleolipids of Canonical Purine ß-d-Ribo-Nucleosides: Synthesis and Cytostatic/Cytotoxic Activities Toward Human and Rat Glioblastoma Cells

We report on the synthesis of two series of canonical purine ß‐d‐ribonucleoside nucleolipids derived from inosine and adenosine, which have been characterized by elemental analyses, electrospray ionization mass spectrometry (ESI MS) as well as by ¹H and ¹³C NMR, and pH‐dependent UV/Vis spectroscopy. A selection of the novel nucleolipids with different lipophilic moieties were first tested on their cytotoxic effect toward human macrophages. Compounds without a significant inhibitory effect on the viability of the macrophages were tested on their cytostatic/cytotoxic effect toward human astrocytoma/oligodendroglioma GOS‐3 cells as well as against the rat malignant neuroectodermal BT4Ca cell line. In order to additionally investigate the potential molecular mechanisms involved in the cytotoxic effects of the derivatives on GOS‐3 or BT4Ca cells, we evaluated the induction of apoptosis and observed the particular activity of the nucleolipid ethyl 3‐{4‐hydroxymethyl‐2‐methyl‐6‐[6‐oxo‐1‐(3,7,11‐trimethyl‐dodeca‐2,6,10‐trienyl)‐1,6‐dihydro‐purin‐9‐yl]‐tetrahydro‐furo[3,4‐d][1,3]dioxol‐2‐yl}propionate (8 c) toward both human and rat glioblastoma cell lines in vitro.

Setup and validation of shake-flask procedures for the determination of partition coefficients (logD) from low drug amounts

Several procedures based on the shake-flask method and designed to require a minimum amount of drug for octanol-water partition coefficient determination have been established and developed. The procedures have been validated by a 28 substance set with a lipophilicity range from -2.0 to 4.5 (logD7.4). The experimental partition is carried out using aqueous phases buffered with phosphate (pH 7.4) and n-octanol saturated with buffered water and the analysis is performed by liquid chromatography. In order to have accurate results, four procedures and eight different ratios between phase volumes are proposed. Each procedure has been designed and optimized (for partition ratios) for a specific range of drug lipophilicity (low, regular and high lipophilicity) and solubility (high and low aqueous solubility). The procedures have been developed to minimize the measurement in the octanolic phase. Experimental logD7.4 values obtained from different procedures and partition ratios show a standard deviation lower than 0.3 and there is a nice agreement when these values are compared with the reference literature ones.

X-3, a mangiferin derivative, stimulates AMP-activated protein kinase and reduces hyperglycemia and obesity in db/db mice

Diabetes mellitus is a major health concern, affecting nearly 10% of the population. Here we describe a potential novel therapeutic agent for this disease, X-3, a derivative of mangiferin. Therapeutic administration of X-3 significantly and dose-dependently reduced plasma glucose and triglycerides in db/db mice following 8 week-treatments. Treatment with X-3 dose-dependently increased the number of insulin-positive β-cell mass. Importantly, X-3 did not cause any death or signs of toxicity in acute toxicity studies. Study of mechanism of action revealed that X-3 increased glucose uptake in parallel with increased phosphorylation of AMP-activated protein kinase (AMPK) in 3T3-L1 cells. It activates AMPK in both LKB1-dependent and -independent manner. Furthermore, administration of X-3 resulted in activation of AMPK and its downstream target, acetyl-CoA carboxylase (ACC) in the hypothalamus, liver, muscle and adipose tissues of C57BL/6 mice. An 80 mg/kg X-3 was more potent than metformin at 500 mg/kg in the hypothalamus, and interscapular fat tissues, potent than MF at the same dose in the liver. Thus, we conclude that X-3 is a promising new class of AMPK activating drug, and can potentially be used in the treatment of type 2 diabetes.

Robustness in experimental design: A study on the reliability of selection approaches

The quality criteria for experimental design approaches in chemoinformatics are numerous. Not only the error performance of a model resulting from the selected compounds is of importance, but also reliability, consistency, stability and robustness against small variations in the dataset or structurally diverse compounds. We developed a new stepwise, adaptive approach, DescRep, combining an iteratively refined descriptor selection with a sampling based on the putatively most representative compounds. A comparison of the proposed strategy was based on statistical performance of models derived from such a selection to those derived by other popular and frequently used approaches, such as the Kennard-Stone algorithm or the most descriptive compound selection. We used three datasets to carry out a statistical evaluation of the performance, reliability and robustness of the resulting models. Our results indicate that stepwise and adaptive approaches have a better adaptability to changes within a dataset and that this adaptability results in a better error performance and stability of the resulting models.

From descriptors to predicted properties: experimental design by using applicability domain estimation

The importance of reliable methods for representative sub-sampling in terms of experimental design and risk assessment within the European Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) system is crucial. We developed experimental design approaches, by utilising predicted properties and the 'distance to model' parameter, to estimate the benefits of certain compounds to the quality of a resulting model. A statistical evaluation of four regression data sets and one classification data set showed that the adaptive concept of iteratively refining the representation of the chemical space contributes to a more efficient and more reliable selection in comparison to traditional approaches. The evaluation of compounds with regard to the uncertainty and the correlation of prediction is beneficial, and in particular, for regression data sets of sufficient size, whereas the use of predicted properties to define the chemical space is beneficial for classification models.

Structural basis and SAR for G007-LK, a lead stage 1,2,4-triazole based specific tankyrase 1/2 inhibitor

Tankyrases 1 and 2 (TNKS1/2) are promising pharmacological biotargets with possible applications for the development of novel anticancer therapeutics. A focused structure-activity relationship study was conducted based on the tankyrase inhibitor JW74 (1). Chemical analoging of 1 improved the 1,2,4-triazole based core and led to 4-{5-[(E)-2-{4-(2-chlorophenyl)-5-[5-(methylsulfonyl)pyridin-2-yl]-4H-1,2,4-triazol-3-yl}ethenyl]-1,3,4-oxadiazol-2-yl}benzonitrile (G007-LK), a potent, "rule of 5" compliant and a metabolically stable TNKS1/2 inhibitor. G007-LK (66) displayed high selectivity toward tankyrases 1 and 2 with biochemical IC50 values of 46 nM and 25 nM, respectively, and a cellular IC50 value of 50 nM combined with an excellent pharmacokinetic profile in mice. The PARP domain of TNKS2 was cocrystallized with 66, and the X-ray structure was determined at 2.8 Å resolution in the space group P3221. The structure revealed that 66 binds to unique structural features in the extended adenosine binding pocket which forms the structural basis for the compound's high target selectivity and specificity. Our study provides a significantly optimized compound for targeting TNKS1/2 in vitro and in vivo.

Web tools for predictive toxicology model building

INTRODUCTION

The development and use of web tools in chemistry has accumulated more than 15 years of history already. Powered by the advances in the Internet technologies, the current generation of web systems are starting to expand into areas, traditional for desktop applications. The web platforms integrate data storage, cheminformatics and data analysis tools. The ease of use and the collaborative potential of the web is compelling, despite the challenges.

AREAS COVERED

The topic of this review is a set of recently published web tools that facilitate predictive toxicology model building. The focus is on software platforms, offering web access to chemical structure-based methods, although some of the frameworks could also provide bioinformatics or hybrid data analysis functionalities. A number of historical and current developments are cited. In order to provide comparable assessment, the following characteristics are considered: support for workflows, descriptor calculations, visualization, modeling algorithms, data management and data sharing capabilities, availability of GUI or programmatic access and implementation details.

EXPERT OPINION

The success of the Web is largely due to its highly decentralized, yet sufficiently interoperable model for information access. The expected future convergence between cheminformatics and bioinformatics databases provides new challenges toward management and analysis of large data sets. The web tools in predictive toxicology will likely continue to evolve toward the right mix of flexibility, performance, scalability, interoperability, sets of unique features offered, friendly user interfaces, programmatic access for advanced users, platform independence, results reproducibility, curation and crowdsourcing utilities, collaborative sharing and secure access.

Online chemical modeling environment (OCHEM): web platform for data storage, model development and publishing of chemical information

The Online Chemical Modeling Environment is a web-based platform that aims to automate and simplify the typical steps required for QSAR modeling. The platform consists of two major subsystems: the database of experimental measurements and the modeling framework. A user-contributed database contains a set of tools for easy input, search and modification of thousands of records. The OCHEM database is based on the wiki principle and focuses primarily on the quality and verifiability of the data. The database is tightly integrated with the modeling framework, which supports all the steps required to create a predictive model: data search, calculation and selection of a vast variety of molecular descriptors, application of machine learning methods, validation, analysis of the model and assessment of the applicability domain. As compared to other similar systems, OCHEM is not intended to re-implement the existing tools or models but rather to invite the original authors to contribute their results, make them publicly available, share them with other users and to become members of the growing research community. Our intention is to make OCHEM a widely used platform to perform the QSPR/QSAR studies online and share it with other users on the Web. The ultimate goal of OCHEM is collecting all possible chemoinformatics tools within one simple, reliable and user-friendly resource. The OCHEM is free for web users and it is available online at http://www.ochem.eu.